Patent Application
20220076709
The present invention relates generally to encoding, transmission and decoding of multimedia files. More specifically, the invention relates to the encoding, transmission and decoding of multimedia files that can include tracks in addition to a single audio track and a single video track.
The development of the internet has prompted the development of file formats for multimedia information to enable standardized generation, distribution and display of multimedia information. Typically, a single multimedia file includes a single video track and a single audio track. When multimedia is written to a high volume and physically transportable medium, such as a CD-R, multiple files can be used to provide a number of video tracks, audio tracks and subtitle tracks. Additional files can be provided containing information that can be used to generate an interactive menu.
Embodiments of the present invention include multimedia files and systems for generating, distributing and decoding multimedia files. In one aspect of the invention, the multimedia files include a plurality of encoded video tracks. In another aspect of the invention, the multimedia files include a plurality of encoded audio tracks. In another aspect of the invention, the multimedia files include at least one subtitle track. In another aspect of the invention, the multimedia files include encoded information indexing the audio, video and/or subtitle tracks.
In one embodiment, the invention includes a series of encoded video frames, a first index that includes information indicative of the location within the file and characteristics of each encoded video frame, a separate second index that includes information indicative of the location within the file of a subset of the encoded video frames.
In a further embodiment, the second index includes at least one tag that references an encoded video frame in the subset of encoded video frames and each tag includes the location within the file of the referenced encoded video frame and the frame number of the encoded video frame in the sequence of encoded video frames.
Another embodiment also includes at least one audio track. In addition, each tag further comprises a reference to a portion of at least one of the audio tracks and the portion of the track that is referenced accompanies the encoded video frame referenced by the tag.
In a still further embodiment, each tag further comprises a reference to information located within the first index and the information referenced in the first index is indicative of the location within the file and characteristics of the encoded video frame referenced by the tag.
Still another embodiment includes a processor and a memory including a file containing at least one sequence of encoded video frames. In addition, the processor is configured to generate an abridged index that references a subset of the encoded video frames in the sequence of encoded video frames.
In a yet further embodiment, the processor is configured to generate a complete index that references all of the encoded video frames in the sequence of encoded video frames and each reference to an encoded video frame in the abridged index includes a reference to the reference to that frame in the complete index.
In yet another embodiment, each reference to an encoded video frame in the abridged index includes the sequence number of the encoded video frame.
In further embodiment again, the processor is configured to include in each reference to an encoded video frame a reference to a location within at least one sound track.
Another embodiment again includes a processor and a memory containing a multimedia file. In addition, the multimedia file includes a sequence of encoded video frames, a complete index referencing each encoded video frame in the sequence of encoded video frames and an abridged index referencing a subset of the encoded video frames in the sequence of encoded video frames. Furthermore, the processor is configured to locate a particular encoded video frame within the multimedia file using the abridged index.
In a further additional embodiment, the processor is configured to locate reference information in the complete index using the abridged index.
In another additional embodiment, the multimedia file includes at least one audio track accompanying the sequence of encoded video frames and each reference to an encoded video frame in the abridged index includes a reference to a portion of at least one of the video tracks.
An embodiment of the method of the invention includes obtaining a sequence of encoded video frames, identifying a subset of frames from the sequence of encoded video frames and generating an abridged index that references the location within the multimedia file of each encoded video frame in the subset of encoded video frames.
In a further embodiment of the method of the invention, the generation of an abridged index further includes generating a reference to the location of each encoded video frame within the subset of encoded video frames and recording the sequence number of each encoded video frame within the subset of encoded video frames.
Another embodiment of the method of the invention also includes obtaining at least one audio track accompanying the sequence of encoded video frames. In addition, the generation of an abridged index further comprises associating with each referenced encoded video frame a reference to a location with at least one of the audio tracks.
A still further embodiment of the method of the invention also includes obtaining a complete index that includes a reference to each encoded video frame in the sequence of encoded video frames. In addition, the generation of an index further comprises associating with each referenced encoded video frame a reference to a location within the second index.
Still another embodiment of the method of the invention includes identifying a desired encoded video frame, determining the encoded video frame that is closest to the desired video frame in the second index and displaying an encoded video frame.
In a yet further embodiment of the method of the invention, each reference in the second index to an encoded video frame also includes a reference to the portion of the first index that refers to that encoded video frame, displaying an encoded video frame, further includes using the reference to the encoded video frame in the second index that is closest to the desired encoded video frame to locate that encoded frame within the first index, searching in the first index for the desired encoded video frame and displaying the desired encoded video frame.
In yet another embodiment of the method of the invention, the closest frame is the closest preceding frame in the sequence to the desired frame.
In a further additional embodiment of the invention, displaying an encoded video frame further includes displaying the encoded video frame that is determined to be closest to the desired video frame.
Referring to the drawings, embodiments of the present invention are capable of encoding, transmitting and decoding multimedia files. Multimedia files in accordance with embodiments of the present invention can contain multiple video tracks, multiple audio tracks, multiple subtitle tracks, a complete index that can be used to locate each data chunk in each of the tracks and an abridged index that can be used to locate a subset of the data chunks in each track, data that can be used to generate a menu interface to access the contents of the file and ‘meta data’ concerning the contents of the file. Multimedia files in accordance with several embodiments of the present invention also include references to video tracks, audio tracks, subtitle tracks and ‘meta data’ external to the file.
Turning now to
In operation, the computer 12 can be used to encode multimedia files in accordance with an embodiment of the present invention. The computer 12 can also be used to decode multimedia files in accordance with embodiments of the present invention and distribute multimedia files in accordance with embodiments of the present invention. The computer can distribute files using any of a variety of file transfer protocols including via a peer-to-peer network. In addition, the computer 12 can transfer multimedia files in accordance with embodiments of the present invention to a server 18, where the files can be accessed by other devices. The other devices can include any variety of computing device or even a dedicated decoder device. In the illustrated embodiment, a set-top cable box, a lap-top computer and a PDA are shown. In other embodiments, various types of digital set-top boxes, desk-top computers, game machines, CE devices and other devices can be connected to the network, download the multimedia files and decode them.
In one embodiment, the devices access the multimedia files from the server via the network. In other embodiments, the devices access the multimedia files from a number of computers via a peer-to-peer network. In several embodiments, multimedia files can be written to a portable storage device such as a disk drive, CD-ROM or DVD. In many embodiments, electronic devices can access multimedia files written to portable storage devices.
Multimedia files in accordance with embodiments of the present invention can be structured to be compliant with the Resource Interchange File Format (‘RIFF file format’), defined by Microsoft Corporation of Redmond, Wash. and International Business Machines Corporation of Armonk, N.Y. RIFF is a file format for storing multimedia data and associated information. A RIFF file typically has an 8-byte RIFF header, which identifies the file and provides the residual length of the file after the header (i.e. file_length−8). The entire remainder of the RIFF file comprises “chunks” and “lists.” Each chunk has an 8-byte chunk header identifying the type of chunk, and giving the length in bytes of the data following the chunk header. Each list has an 8-byte list header identifying the type of list and giving the length in bytes of the data following the list header. The data in a list comprises chunks and/or other lists (which in turn may comprise chunks and/or other lists). RIFF lists are also sometimes referred to as “list chunks.”
An AVI file is a special form of RIFF file that follow the format of a RIFF file, but include various chunks and lists with defined identifiers that contain multimedia data in particular formats. The AVI format was developed and defined by Microsoft Corporation. AVI files are typically created using a encoder that can output multimedia data in the AVI format. AVI files are typically decoded by any of a group of software collectively known as AVI decoders.
The RIFF and AVI formats are flexible in that they only define chunks and lists that are part of the defined file format, but allow files to also include lists and/or chunks that are outside the RIFF and/or AVI file format definitions without rendering the file unreadable by a RIFF and/or AVI decoder. In practice, AVI (and similarly RIFF) decoders are implemented so that they simply ignore lists and chunks that contain header information not found in the AVI file format definition. The AVI decoder must still read through these non-AVI chunks and lists and so the operation of the AVI decoder may be slowed, but otherwise, they generally have no effect on and are ignored by an AVI decoder.
A multimedia file in accordance with an embodiment of the present invention is illustrated in
Another multimedia file in accordance with an embodiment of the present invention is shown in
In the illustrated embodiment, the multimedia file 30′ includes multiple concatenated ‘RIFF’ chunks, where the first ‘RIFF’ chunk 50 includes a character set chunk (‘CSET’ chunk) 32′, an information list chunk (‘INFO’ list chunk) 34′, a file header chunk (‘hdrl’ list chunk) 36′, a meta data chunk (‘DXDT’ chunk) 38′, a menu chunk (‘DMNU’ chunk) 40′, a junk chunk (‘junk’ chunk) 41′, the movie list chunk (‘movi’ list chunk) 42′ and an optional index chunk (‘idx1’ chunk) 44′. The second ‘RIFF’ chunk 52 contains a second menu chunk (‘DMNU’ chunk) 46′. Additional ‘RIFF’ chunks 54 containing additional titles can be included after the ‘RIFF’ menu chunk 52. The additional ‘RIFF’ chunks can contain independent media in compliant AVI file format. In one embodiment, the second menu chunk 46′ and the additional ‘RIFF’ chunks have specialized 4 character codes (defined in the AVI format and discussed below) such that the first two characters of the 4 character codes appear as the second two characters and the second two characters of the 4 character codes appear as the first two characters.
The ‘CSET’ chunk 32 is a chunk defined in the Audio Video Interleave file format (AVI file format), created by Microsoft Corporation. The ‘CSET’ chunk defines the character set and language information of the multimedia file. Inclusion of a ‘CSET’ chunk in accordance with embodiments of the present invention is optional.
A multimedia file in accordance with one embodiment of the present invention does not use the ‘CSET’ chunk and uses UTF-8, which is defined by the Unicode Consortium, for the character set by default combined with RFC 3066 Language Specification, which is defined by Internet Engineering Task Force for the language information.
The ‘INFO’ list chunk 34 can store information that helps identify the contents of the multimedia file. The ‘INFO’ list is defined in the AVI file format and its inclusion in a multimedia file in accordance with embodiments of the present invention is optional. Many embodiments that include a ‘DXDT’ chunk do not include an ‘INFO’ list chunk.
2.3. The ‘hdrl’ List Chunk
The ‘hdrl’ list chunk 38 is defined in the AVI file format and provides information concerning the format of the data in the multimedia file. Inclusion of a ‘hdrl’ list chunk or a chunk containing similar description information is generally required. The ‘hdrl’ list chunk includes a chunk for each video track, each audio track and each subtitle track.
A conceptual diagram of a ‘hdrl’ list chunk 38 in accordance with one embodiment of the invention that includes a single video track 62, two audio tracks 64, an external audio track 66, two subtitle tracks 68 and an external subtitle track 70 is illustrated in
In addition to the ‘avih’ chunk, the ‘hdrl’ list includes a stream descriptor list for each audio, video and subtitle track. In one embodiment, the stream descriptor list is implemented using ‘strl’ chunks. A ‘strl’ chunk in accordance with an embodiment of the present invention is illustrated in
Multimedia files in accordance with embodiments of the present invention can contain references to external files holding multimedia information such as an additional audio track or an additional subtitle track. The references to these tracks can either be contained in the ‘hdrl’ chunk or in the ‘junk’ chunk 41. In either case, the reference can be contained in the ‘strh’ chunk 92 of a ‘strl’ chunk 90, which references either a local file or a file stored remotely. The referenced file can be a standard AVI file or a multimedia file in accordance with an embodiment of the present invention containing the additional track.
In additional embodiments, the referenced file can contain any of the chunks that can be present in the referencing file including ‘DMNU’ chunks, ‘DXDT’ chunks and chunks associated with audio, video and/or subtitle tracks for a multimedia presentation. For example, a first multimedia file could include a ‘DMNU’ chunk (discussed in more detail below) that references a first multimedia presentation located within the ‘movi’ list chunk of the first multimedia file and a second multimedia presentation within the ‘movi’ list chunk of a second multimedia file. Alternatively, both ‘movi’ list chunks can be included in the same multimedia file, which need not be the same file as the file in which the ‘DMNU’ chunk is located.
The ‘DXDT’ chunk 38 contains so called ‘meta data’. ‘Meta data’ is a term used to describe data that provides information about the contents of a file, document or broadcast. The ‘meta data’ stored within the ‘DXDT’ chunk of multimedia files in accordance with embodiments of the present invention can be used to store such content specific information as title, author, copyright holder and cast. In addition, technical details about the codec used to encode the multimedia file can be provided such as the CLI options used and the quantizer distribution after each pass.
In one embodiment, the meta data is represented within the ‘DXDT’ chunk as a series of statements, where each statement includes a subject, a predicate, an object and an authority. The subject is a reference to what is being described. The subject can reference a file, item, person or organization. The subject can reference anything having characteristics capable of description. The predicate identifies a characteristic of the subject that is being described. The object is a description of the identified characteristic of the subject and the authority identifies the source of the information.
The following is a table showing an example of how various pieces of ‘meta data’, can be represented as an object, a predicate, a subject and an authority:
In one embodiment, the expression of the subject, predicate, object and authority is implemented using binary representations of the data, which can be considered to form Directed-Labeled Graphs (DLGs). A DLG consists of nodes that are either resources or literals. Resources are identifiers, which can either be conformant to a naming convention such as a Universal Resource Identifier (“URI”) as defined in RFC 2396 by the Internet Engineering Taskforce (http://www.ietf.org/rfc/rfc2396.txt) or refer to data specific to the system itself. Literals are representations of an actual value, rather than a reference.
An advantage of DLGs is that they allow the inclusion of a flexible number of items of data that are of the same type, such as cast members of a movie. In the example shown in Table 1, three cast members are included. However, any number of cast members can be included. DLGs also allow relational connections to other data types. In Table 1, there is a ‘meta data’ item that has a subject “_:file281,” a predicate “Series,” and an object “_:file321.” The subject “_:file281” indicates that the ‘meta data’ refers to the content of the file referenced as “_:file321” (in this case, a movie—“Movie Title 1”). The predicate is “Series,” indicating that the object will have information about another movie in the series to which the first movie belongs. However, “_:file321” is not the title or any other specific information about the series, but rather a reference to another entry that provides more information about “_:file321”. The next ‘meta data’ entry, with the subject “_:file321”, however, includes data about “_:file321,” namely that the Title as specified by the Dublin Core Vocabulary as indicated by “http://purl.org/dc/elements/1.1/title” of this sequel is “Movie Title 2.”
Additional ‘meta data’ statements in Table 1 specify that “Actor 1” was a member of the cast playing the role of “Character Name 1” and that there are two directors. Technical information is also expressed in the ‘meta data.’ The ‘meta data’ statements identify that “_:file281” includes track “_:track#dc00.” The ‘meta data’ provides information including the resolution of the video track, the certification level of the video track and the codec settings. Although not shown in Table 1, the ‘meta data’ can also include a unique identifier assigned to a track at the time of encoding. When unique identifiers are used, encoding the same content multiple times will result in a different identifier for each encoded version of the content. However, a copy of the encoded video track would retain the identifier of the track from which it was copied.
The entries shown in Table 1 can be substituted with other vocabularies such as the UPnP vocabulary, which is defined by the UPnP forum (see http://www.upnpforum.org). Another alternative would be the Digital Item Declaration Language (DIDL) or DIDL-Lite vocabularies developed by the International Standards Organization as part of work towards the MPEG-21 standard. The following are examples of predicates within the UPnP vocabulary:
urn:schemas-upnp-org:metadata-1-0/upnp/artist
urn:schemas-upnp-org:metadata-1-0/upnp/actor
urn:schemas-upnp-org:metadata-1-0/upnp/author
urn:schemas-upnp-org:metadata-1-0/upnp/producer
urn:schemas-upnp-org:metadata-1-0/upnp/director
urn:schemas-upnp-org:metadata-1-0/upnp/genre
urn:schemas-upnp-org:metadata-1-0/upnp/album
urn:schemas-upnp-org:metadata-1-0/upnp/playlist
urn:schemas-upnp-org:metadata-1-0/upnp/originalTrackNumber
urn:schemas-upnp-org:metadata-1-0/upnp/userAnnotation
The authority for all of the ‘meta data’ is ‘_:auth42.’ ‘Meta data’ statements show that ‘_:auth42’ is ‘Studio Name.’ The authority enables the evaluation of both the quality of the file and the ‘meta data’ statements associated with the file.
Nodes into a graph are connected via named resource nodes. A statement of ‘meta data’ consist of a subject node, a predicate node and an object node. Optionally, an authority node can be connected to the DLG as part of the ‘meta data’ statement.
For each node, there are certain characteristics that help further explain the functionality of the node. The possible types can be represented as follows using the ANSI C programming language:
An example of a data structure (represented in the ANSI C programming language) that represents the ‘meta data’ chunks contained within the ‘DXDT’ chunk is as follows:
The ‘RDFData’ chunk includes a chunk referred to as an ‘RDFHeader’ chunk, a value ‘numOfStatements’ and a list of ‘RDFStatement’ chunks.
The ‘RDFHeader’ chunk contains information about the manner in which the ‘meta data’ is formatted in the chunk. In one embodiment, the data in the ‘RDFHeader’ chunk can be represented as follows (represented in ANSI C):
The ‘RDFHeader’ chunk includes a number ‘version’ that indicates the version of the resource description format to enable forward compatibility. The header includes a second number ‘numOfSchemas’ that represents the number of ‘RDFSchema’ chunks in the list ‘schemas’, which also forms part of the ‘RDFHeader’ chunk. In several embodiments, the ‘RDFSchema’ chunks are used to enable complex resources to be represented more efficiently. In one embodiment, the data contained in a ‘RDFSchema’ chunk can be represented as follows (represented in ANSI C):
The ‘RDFSchema’ chunk includes a first string of text such as ‘dc’ identified as ‘prefix’ and a second string of text such as ‘http://purl.org/dc/elements/1.1/’ identified as ‘uri’. The ‘prefix’ defines a term that can be used in the ‘meta data’ in place of the cuff. The ‘uri’ is a Universal Resource Identifier, which can conform to a specified standardized vocabulary or be a specific vocabulary to a particular system.
Returning to the discussion of the ‘RDFData’ chunk. In addition to a ‘RDFHeader’ chunk, the ‘RDFData’ chunk also includes a value ‘numOfStatements’ and a list ‘statement’ of ‘RDFStatement’ chunks. The value ‘numOfStatements’ indicates the actual number of ‘RDFStatement’ chunks in the list ‘statements’ that contain information. In one embodiment, the data contained in the ‘RDFStatement’ chunk can be represented as follows (represented in ANSI C):
Each ‘RDFStatement’ chunk contains a piece of ‘meta data’ concerning the multimedia file. The chunks ‘subject’, ‘predicate’, ‘object’ and ‘authority’ are used to contain the various components of the ‘meta data’ described above.
The ‘subject’ is a ‘RDFSubject’ chunk, which represents the subject portion of the ‘meta data’ described above. In one embodiment the data contained within the ‘RDFSubject’ chunk can be represented as follows (represented in ANSI C):
The ‘RDFSubject’ chunk shown above includes a value ‘type’ that indicates that the data is either a Resource or an anonymous node of a piece of ‘meta data’ and a unicode text string ‘value’, which contains data representing the subject of the piece of ‘meta data’. In embodiments where an ‘RDFSchema’ chunk has been defined the value can be a defined term instead of a direct reference to a resource.
The ‘predicate’ in a ‘RDFStatement’ chunk is a ‘RDFPredicate’ chunk, which represents the predicate portion of a piece of ‘meta data’. In one embodiment the data contained within a ‘RDFPredicate’ chunk can be represented as follows (represented in ANSI C):
The ‘RDFPredicate’ chunk shown above includes a value ‘type’ that indicates that the data is the predicate URI or an ordinal list entry of a piece of ‘meta data’ and a text string ‘value,’ which contains data representing the predicate of a piece of ‘meta data.’ In embodiments where an ‘RDFSchema’ chunk has been defined the value can be a defined term instead of a direct reference to a resource.
The ‘object’ in a ‘RDFStatement’ chunk is a ‘RDFObject’ chunk, which represents the object portion of a piece of ‘meta data.’ In one embodiment, the data contained in the ‘RDFObject’ chunk can be represented as follows (represented in ANSI C):
The ‘RDFObject’ chunk shown above includes a value ‘type’ that indicates that the piece of data is a UTF-8 literal string, a literal integer or literal XML data of a piece of ‘meta data.’ The chunk also includes three values. The first value ‘language’ is used to represent the language in which the piece of ‘meta data’ is expressed (e.g. a film's title may vary in different languages). In several embodiments, a standard representation can be used to identify the language (such as RFC 3066—Tags for the Identification of Languages specified by the Internet Engineering Task Force, see http://www.ietf.org/rfc/rfc3066.txt). The second value ‘cdataTypeURI’ is used to indicate the type of data that is contained within the ‘value’ field if it can not be explicitly indicated by the ‘type’ field. The URI specified by the dataTypeURI points to general RDF URI Vocabulary used to describe the particular type of the Data is used. Different formats in which the URI can be expressed are described at http://www.w3.org/TR/rdf-concepts/#section-Datatypes. In one embodiment, the ‘value’ is a ‘wide character.’ In other embodiments, the ‘value’ can be any of a variety of types of data from a single bit, to an image or a video sequence. The ‘value’ contains the object piece of the ‘meta data.’
The ‘authority’ in a ‘RDFStatement’ chunk is a ‘RDFAuthority’ chunk, which represents the authority portion of a piece of ‘meta data.’ In one embodiment the data contained within the ‘RDFAuthority’ chunk can be represented as follows (represented in ANSI C):
The ‘RDFAuthority’ data structure shown above includes a value ‘type’ that indicates the data is a Resource or an anonymous node of a piece of ‘meta data.’ The ‘value’ contains the data representing the authority for the ‘meta data.’ In embodiments where an ‘RDFSchema’ chunk has been defined the value can be a defined term instead of a direct reference to a resource.
A conceptual representation of the storage of a ‘DXDT’ chunk of a multimedia file in accordance with an embodiment of the present invention is shown in
As is discussed below, multimedia files in accordance with embodiments of the present invention can be continuously modified and updated. Determining in advance the ‘meta data’ to associate with the file itself and the ‘meta data’ to access remotely (e.g. via the internet) can be difficult. Typically, sufficient ‘meta data’ is contained within a multimedia file in accordance with an embodiment of the present invention in order to describe the contents of the file. Additional information can be obtained if the device reviewing the file is capable of accessing via a network other devices containing ‘meta data’ referenced from within the file.
The methods of representing ‘meta data’ described above can be extendable and can provide the ability to add and remove different ‘meta data’ fields stored within the file as the need for it changes over time. In addition, the representation of ‘meta data’ can be forward compatible between revisions.
The structured manner in which ‘meta data’ is represented in accordance with embodiments of the present invention enables devices to query the multimedia file to better determine its contents. The query could then be used to update the contents of the multimedia file, to obtain additional ‘meta data’ concerning the multimedia file, generate a menu relating to the contents of the file or perform any other function involving the automatic processing of data represented in a standard format. In addition, defining the length of each parseable element of the ‘meta data’ can increase the ease with which devices with limited amounts of memory, such as consumer electronics devices, can access the ‘meta data’.
In other embodiments, the ‘meta data’ is represented using individual chunks for each piece of ‘meta data.’ Several ‘DXDT’ chunks in accordance with the present invention include a binary chunk containing ‘meta data’ encoded as described above and additional chunks containing individual pieces of ‘meta data’ formatted either as described above or in another format. In embodiments where binary ‘meta data’ is included in the ‘DXDT’ chunk, the binary ‘meta data’ can be represented using 64-bit encoded ASCII. In other embodiments, other binary representations can be used.
Examples of individual chunks that can be included in the ‘DXDT’ chunk in accordance with the present invention are illustrated in
The ‘PixelAspectRatioMetaData’ chunk 152a includes information concerning the pixel aspect ratio of the encoded video. The ‘EncoderURIMetaData’ chunk 152b includes information concerning the encoder. The ‘CodecSettingsMetaData’ chunk 152c includes information concerning the settings of the codec used to encode the video. The ‘FrameTypeMetaData’ chunk 152d includes information concerning the video frames. The ‘VideoResolutionMetaData’ chunk 152e includes information concerning the video resolution of the encoded video. The ‘PublisherMetaData’ chunk 152f includes information concerning the person or organization that published the media. The ‘CreatorMetaData’ chunk 152g includes information concerning the creator of the content. The ‘GenreMetaData’ chunk 152h includes information concerning the genre of the media. The ‘CreatorToolMetaData’ chunk 152i includes information concerning the tool used to create the file. The ‘RightsMetaData’ chunk 152j includes information concerning DRM. The ‘RunTimeMetaData’ chunk 152k includes information concerning the run time of the media. The ‘QuantizerMetaData’ chunk 152l includes information concerning the quantizer used to encode the video. The ‘CodecInfoMetaData’ chunk 152m includes information concerning the codec. The ‘EncoderNameMetaData’ chunk 152n includes information concerning the name of the encoder. The ‘FrameRateMetaData’ chunk 152o includes information concerning the frame rate of the media. The ‘inputSourceMetaData’ chunk 152p includes information concerning the input source. The ‘FileIDMetaData’ chunk 152q includes a unique identifier for the file. The ‘TypeMetaData’ chunk 152r includes information concerning the type of the multimedia file. The ‘TitleMetaData’ chunk 152s includes the title of the media and the ‘CertLevelMetaData’ chunk 152t includes information concerning the certification level of the media. In other embodiments, additional chunks can be included that contain additional ‘meta data.’ In several embodiments, a chunk containing ‘meta data’ in a binary format as described above can be included within the ‘MetaData’ chunk. In one embodiment, the chunk of binary ‘meta data’ is encoded as 64-bit ASCII.
In one embodiment, the ‘DXDT’ chunk can include an ‘index’ chunk that can be used to index ‘data’ chunks in the ‘movi’ list chunk 42 corresponding to specific frames in a sequence of encoded video (the ‘movi’ list chunk is discussed further below). The ‘index’ chunk can be differentiated from the ‘idx1’ chunk on the basis that the ‘index’ chunk does not include information concerning every ‘data’ chunk in the ‘movi’ list chunk. Typically, the ‘index’ chunk includes information concerning a subset of the ‘data’ chunks. Appropriate selection of the ‘data’ chunks referenced in the ‘index’ chunk can enable rapid location of a specific video frame.
An embodiment of an ‘index’ chunk in accordance with the present invention is shown in
In the illustrated embodiment, each ‘tag’ chunk 154 includes information that can be used to reference a specific encoded video frame and the portions of one or more audio tracks that accompany the encoded video frame. The information includes information concerning a chunk offset 155a, information concerning an index offset 155b, information identifying a video frame 155c and information identifying a portion of an audio track 155d.
In many embodiments, the chunk offset 155a can be used to locate the ‘data’ chunks within the multimedia file corresponding to the particular encoded video frame and/or accompanying audio track(s) referenced by the ‘tag’ chunk. In one embodiment, the chunk offset is the the location of the relevant ‘data’ chunk relative to the start of the ‘movi’ list chunk.
In many embodiments, the index offset 155b can be used to locate information about a particular video frame in the ‘idx1’ chunk. In one embodiment, the index offset 155b is the location of the relevant piece of information relative to the beginning of the ‘idx1’ chunk.
The information identifying the video frame 155c designates the position of the encoded video frame referenced by the ‘tag’ chunk within a sequence of encoded video frames. In one embodiment, the information 155c can simply be an integer indicating a frame number in a sequence of encoded video frames. In other embodiments, other information indicative of the position of the encoded video frame referenced by the ‘tag’ chunk within the video sequence can be used.
In many embodiments, the information identifying a portion of an audio track 155d references a specific section of an audio track. In several embodiments, the identified section corresponds to the portion of a soundtrack that accompanies the encoded video frame referenced by the ‘tag’ chunk. In embodiments where there are multiple audio tracks, the information identifying a portion of an audio track 155d can in fact include multiple pieces of information capable of referencing sections within each of the multiple audio tracks. In many embodiments, the sections identified by the multiple pieces of information correspond to the portion of each sound track that accompanies an encoded video frame referenced by the ‘tag’ chunk.
In many embodiments of ‘tag’ chunks in accordance with the present invention, various pieces of information are used to identify the ‘data’ chunk within a multimedia file corresponding to a specific encoded video frame. In addition, various types of information can be used to identify portions of audio and subtitle tracks that accompany an encoded video frame. In several embodiments, at least some of the ‘tag’ chunks in an ‘index’ chunk reference a portion of an audio track and do not reference an encoded video frame.
Including a chunk containing index information before the ‘movi’ list chunk can enable a device to start playing and performing other functions, such as fast forward, rewind and scene skipping, prior to the downloading of the ‘idx1’ chunk. In one embodiment, the ‘index’ chunk is included in a chunk other than the ‘DXDT’ chunk preceding the ‘movi’ list chunk (e.g. the junk chunk). In other embodiments, the ‘index’ chunk is not located within the ‘junk’ chunk. In several embodiments, the ‘index’ chunk is a separate chunk. In one embodiment, the ‘index’ chunk is located after the ‘movi’ list chunk.
Referring to
The ability to provide a so-called ‘lite’ menu can be useful for consumer electronics devices that cannot process the amounts of data required for more sophisticated menu systems. Providing a menu (whether ‘lite’ or otherwise) prior to the ‘movi’ list chunk 42 can reduce delays when playing embodiments of multimedia files in accordance with the present invention in streaming or progressive download applications. In several embodiments, providing a simple and a complex menu can enable a device to choose the menu that it wishes to display. Placing the smaller of the two menus before the ‘movi’ list chunk 42 enables devices in accordance with embodiments of the present invention that cannot display menus to rapidly skip over information that cannot be displayed.
In other embodiments, the data required to create a single menu is split between the first and second ‘DMNU’ chunks. Alternatively, the ‘DMNU’ chunk can be a single chunk before the ‘movi’ chunk containing data for a single set of menus or multiple sets of menus. In other embodiments, the ‘DMNU’ chunk can be a single or multiple chunks located in other locations throughout the multimedia file.
In several multimedia files in accordance with the present invention, the first ‘DMNU’ chunk 40 (40′) can be automatically generated based on a ‘richer’ menu in the second ‘DMNU’ chunk 46 (46′). The automatic generation of menus is discussed in greater detail below.
The structure of a ‘DMNU’ chunk in accordance with an embodiment of the present invention is shown in
In one embodiment, the ‘WowMenu’ chunk 160 contains the hierarchy of menu chunk objects that are conceptually illustrated in
Use of ‘LanguageMenus’ chunks 172 enables the ‘DMNU’ chunk 158 to contain menu information in different languages. Each ‘LanguageMenus’ chunk 172 contains the information used to generate a complete set of menus in a specified language. Therefore, the ‘LanguageMenus’ chunk includes an identifier that identifies the language of the information associated with the ‘LanguageMenus’ chunk. The ‘LanguageMenus’ chunk also includes a list of ‘WowMenu’ chunks 175.
Each ‘WowMenu’ chunk 175 contains all of the information to be displayed on the screen for a particular menu. This information can include background video and audio. The information can also include data concerning button actions that can be used to access other menus or to exit the menu and commence displaying a portion of the multimedia file. In one embodiment, the ‘WowMenu’ chunk 175 includes a list of references to media. These references refer to information contained in the ‘Media’ chunk 174, which will be discussed further below. The references to media can define the background video and background audio for a menu. The ‘WowMenu’ chunk 175 also defines an overlay that can be used to highlight a specific button, when a menu is first accessed.
In addition, each ‘WowMenu’ chunk 175 includes a number of ‘ButtonMenu’ chunks 176. Each ‘ButtonMenu’ chunk defines the properties of an onscreen button. The ‘ButtonMenu’ chunk can describe such things as the overlay to use when the button is highlighted by the user, the name of the button and what to do in response to various actions performed by a user navigating through the menu. The responses to actions are defined by referencing an ‘Action’ chunk 178. A single action, e.g. selecting a button, can result in several ‘Action’ chunks being accessed. In embodiments where the user is capable of interacting with the menu using a device such as a mouse that enables an on-screen pointer to move around the display in an unconstrained manner, the on-screen location of the buttons can be defined using a ‘MenuRectangle’ chunk 180. Knowledge of the on-screen location of the button enables a system to determine whether a user is selecting a button, when using a free ranging input device.
Each ‘Action’ chunk identifies one or more of a number of different varieties of action related chunks, which can include a ‘PlayAction’ chunk 182, a ‘MenuTransitionAction’ chunk 184, a ‘ReturnToPlayAction’ chunk 186, an ‘AudioSelectAction’ chunk 188, a ‘SubtitileSelectAction’ chunk 190 and a ‘ButtonTransitionAction’ chunk 191. A ‘PlayAction’ chunk 182 identifies a portion of each of the video, audio and subtitle tracks within a multimedia file. The ‘PlayAction’ chunk references a portion of the video track using a reference to a ‘MediaTrack’ chunk (see discussion below). The ‘PlayAction’ chunk identifies audio and subtitle tracks using ‘SubtitleTrack’ 192 and ‘AudioTrack’ 194 chunks. The ‘SubtitleTrack’ and ‘AudioTrack’ chunks both contain references to a ‘MediaTrack’ chunk 198. When a ‘PlayAction’ chunk forms the basis of an action in accordance with embodiments of the present invention, the audio and subtitle tracks that are selected are determined by the values of variables set initially as defaults and then potentially modified by a user's interactions with the menu.
Each ‘MenuTransitionAction’ chunk 184 contains a reference to a ‘WowMenu’ chunk 175. This reference can be used to obtain information to transition to and display another menu.
Each ‘ReturnToPlayAction’ chunk 186 contains information enabling a player to return to a portion of the multimedia file that was being accessed prior to the user bringing up a menu.
Each ‘AudioSelectAction’ chunk 188 contains information that can be used to select a particular audio track. In one embodiment, the audio track is selected from audio tracks contained within a multimedia file in accordance with an embodiment of the present invention. In other embodiments, the audio track can be located in an externally referenced file.
Each ‘SubtitleSelectAction’ chunk 190 contains information that can be used to select a particular subtitle track. In one embodiment, the subtitle track is selected from a subtitle contained within a multimedia file in accordance with an embodiment of the present invention. In other embodiments, the subtitle track can be located in an externally referenced file.
Each ‘ButtonTransitionAction’ chunk 191 contains information that can be used to transition to another button in the same menu. This is performed after other actions associated with a button have been performed.
The ‘Media’ chunk 174 includes a number of ‘MediaSource’ chunks 166 and ‘MediaTrack’ chunks 198. The ‘Media’ chunk defines all of the multimedia tracks (e.g., audio, video, subtitle) used by the feature and the menu system. Each ‘MediaSource’ chunk 196 identifies a RIFF chunk within the multimedia file in accordance with an embodiment of the present invention, which, in turn, can include multiple RIFF chunks.
Each ‘MediaTrack’ chunk 198 identifies a portion of a multimedia track within a RIFF chunk specified by a ‘MediaSource’ chunk.
The ‘MRIF’ chunk 162 is, essentially, its own small multimedia file that complies with the RIFF format. The ‘MRIF’ chunk contains audio, video and subtitle tracks that can be used to provide background audio and video and overlays for menus. The ‘MRIF’ chunk can also contain video to be used as overlays to indicate highlighted menu buttons. In embodiments where less menu data is required, the background video can be a still frame (a variation of the AVI format) or a small sequence of identical frames. In other embodiments, more elaborate sequences of video can be used to provide the background video.
As discussed above, the various chunks that form part of a ‘WowMenu’ chunk 175 and the ‘WowMenu’ chunk itself contain references to actual media tracks. Each of these references is typically to a media track defined in the ‘hdrl’ LIST chunk of a RIFF chunk.
Other chunks that can be used to create a ‘DMNU’ chunk in accordance with the present invention are shown in
The contents of the ‘LanguageMenus’ chunk 172′ is largely similar to that of the ‘LanguageMenus’ chunk 172 illustrated in
The ‘Media’ chunk 174′ is significantly different from the ‘Media’ chunk 174 shown in
The ‘Title’ chunk can contain at least one ‘Chapter’ chunk 206. The ‘Chapter’ chunk 206 references a scene within a particular title. The ‘Chapter’ chunk 206 contains references to the portions of the video track, each audio track and each subtitle track that correspond to the scene indicated by the ‘Chapter’ chunk. In one embodiment, the references are implemented using ‘MediaSource’ chunks 196′ and ‘MediaTrack’ chunks 198′ similar to those described above in relation to
As described above, the ‘MenuTracks’ chunks 204 contain references to the media that are used to generate the audio, video and overlay media of the menus. In one embodiment, the references to the media information are made using ‘MediaSource’ chunks 196′ and ‘MediaTrack’ chunks 198′ contained within the ‘MenuTracks’ chunk. In one embodiment, the ‘MediaSource’ chunks 196′ and ‘MediaTrack’ chunks 198′ are implemented in the manner described above in relation to
The ‘TranslationTable’ chunk 200 can be used to contain text strings describing each title and chapter in a variety of languages. In one embodiment, the ‘TranslationTable’ chunk 200 includes at least one ‘TranslationLookup’ chunk 208. Each ‘TranslationLookup’ chunk 208 is associated with a ‘Title’ chunk 202, a ‘Chapter’ chunk 206 or a ‘MediaTrack’ chunk 196′ and contains a number of ‘Translation’ chunks 210. Each of the ‘Translation’ chunks in a ‘TranslationLookup’ chunk contains a text string that describes the chunk associated with the ‘TranslationLookup’ chunk in a language indicated by the ‘Translation’ chunk.
A diagram conceptually illustrating the relationships between the various chunks contained within a ‘DMNU’ chunk is illustrated in
2.6. The ‘junk’ Chunk
The ‘junk’ chunk 41 is an optional chunk that can be included in multimedia files in accordance with embodiments of the present invention. The nature of the ‘junk’ chunk is specified in the AVI file format.
2.7. The ‘movi’ List Chunk
The ‘movi’ list chunk 42 contains a number of ‘data’ chunks. Examples of information that ‘data’ chunks can contain are audio, video or subtitle data. In one embodiment, the ‘movi’ list chunk includes data for at least one video track, multiple audio tracks and multiple subtitle tracks.
The interleaving of ‘data’ chunks in the ‘movi’ list chunk 42 of a multimedia file containing a video track, three audio tracks and three subtitle tracks is illustrated in
Adjacent ‘video’ chunks may also be separated by one or more ‘subtitle’ chunks 266 from one of the subtitle tracks. In one embodiment, the ‘subtitle’ chunk 266 includes a subtitle and a start time and a stop time. In several embodiments, the ‘subtitle’ chunk is interleaved in the ‘movi’ list chunk such that the ‘video’ chunk following the ‘subtitle’ chunk includes the portion of video that occurs at the start time of the subtitle. In other embodiments, the start time of all ‘subtitle’ and ‘audio’ chunks is ahead of the equivalent start time of the video. In one embodiment, the ‘audio’ and ‘subtitle’ chunks can be placed within 5 seconds of the corresponding ‘video’ chunk and in other embodiments the ‘audio’ and ‘subtitle’ chunks can be placed within a time related to the amount of video capable of being buffered by a device capable of displaying the audio and video within the file.
In one embodiment, the ‘data’ chunks include a ‘FOURCC’ code to identify the stream to which the ‘data’ chunk belongs. The ‘FOURCC’ code consists of a two-digit stream number followed by a two-character code that defines the type of information in the chunk. An alternate ‘FOURCC’ code consists of a two-character code that defines the type of information in the chunk followed by the two-digit stream number. Examples of the two-character code are shown in the following table:
In one embodiment, the structure of the ‘video’ chunks 262 and ‘audio’ chunks 264 complies with the AVI file format. In other embodiments, other formats for the chunks can be used that specify the nature of the media and contain the encoded media.
In several embodiments, the data contained within a ‘subtitle’ chunk 266 can be represented as follows:
The value ‘fcc’ is the FOURCC code that indicates the subtitle track and nature of the subtitle track (text or bitmap mode). The value ‘cb’ specifies the size of the structure. The value ‘duration’ specifies the time at the starting and ending point of the subtitle. In one embodiment, it can be in the form hh:mm:ss.xxx-hh:mm:ss.xxx. The hh represent the hours, mm the minutes, ss the seconds and xxx the milliseconds. The value ‘subtitle’ contains either the Unicode text of the subtitle in text mode or a bitmap image of the subtitle in the bitmap mode. Several embodiments of the present invention use compressed bitmap images to represent the subtitle information. In one embodiment, the ‘subtitle’ field contains information concerning the width, height and onscreen position of the subtitle. In addition, the ‘subtitle’ field can also contain color information and the actual pixels of the bit map. In several embodiments, run length coding is used to reduce the amount of pixel information required to represent the bitmap.
Multimedia files in accordance with embodiments of the present invention can include digital rights management. This information can be used in video on demand applications. Multimedia files that are protected by digital rights management can only be played back correctly on a player that has been granted the specific right of playback. In one embodiment, the fact that a track is protected by digital rights management can be indicated in the information about the track in the ‘hdrl’ list chunk (see description above). A multimedia file in accordance with an embodiment of the present invention that includes a track protected by digital rights management can also contain information about the digital rights management in the ‘movi’ list chunk.
A ‘movi’ list chunk of a multimedia file in accordance with an embodiment of the present invention that includes a video track, multiple audio tracks, at least one subtitle track and information enabling digital rights management is illustrated in
In an encryption system in accordance with an embodiment of the present invention, the video chunks are only partially encrypted. Where partial encryption is used, the ‘DRM’ chunks contain a reference to the portion of a ‘video’ chunk that is encrypted and a reference to the key that can be used to decrypt the encrypted portion. The decryption keys can be located in a ‘DRM’ header, which is part of the ‘strd’ chunk (see description above). The decryption keys are scrambled and encrypted with a master key. The ‘DRM’ header also contains information identifying the master key.
A conceptual representation of the information in a ‘DRM’ chunk is shown in
2.8. The ‘idx1’ Chunk
The ‘idx1’ chunk 44 is an optional chunk that can be used to index the ‘data’ chunks in the ‘movi’ list chunk 42. In one embodiment, the ‘idx1’ chunk can be implemented as specified in the AVI format. In other embodiments, the ‘idx1’ chunk can be implemented using data structures that reference the location within the file of each of the ‘data’ chunks in the ‘movi’ list chunk. In several embodiments, the ‘idx1’ chunk identifies each ‘data’ chunk by the track number of the data and the type of the data. The FOURCC codes referred to above can be used for this purpose.
Embodiments of the present invention can be used to generate multimedia files in a number of ways. In one instance, systems in accordance with embodiments of the present invention can generate multimedia files from files containing separate video tracks, audio tracks and subtitle tracks. In such instances, other information such as menu information and ‘meta data’ can be authored and inserted into the file.
Other systems in accordance with embodiments of the present invention can be used to extract information from a number of files and author a single multimedia file in accordance with an embodiment of the present invention. Where a CD-R is the initial source of the information, systems in accordance with embodiments of the present invention can use a codec to obtain greater compression and can re-chunk the audio so that the audio chunks correspond to the video chunks in the newly created multimedia file. In addition, any menu information in the CD-R can be parsed and used to generate menu information included in the multimedia file.
Other embodiments can generate a new multimedia file by adding additional content to an existing multimedia file in accordance with an embodiment of the present invention. An example of adding additional content would be to add an additional audio track to the file such as an audio track containing commentary (e.g. director's comments, after-created narrative of a vacation video). The additional audio track information interleaved into the multimedia file could also be accompanied by a modification of the menu information in the multimedia file to enable the playing of the new audio track.
A system in accordance with an embodiment of the present invention for generating a multimedia file is illustrated in
Typically the chunks provided to the interleaver are stored on a storage device. In several embodiments, all of the chunks are stored on the same storage device. In other embodiments, the chunks may be provided to the interleaver from a variety of storage devices or generated and provided to the interleaver in real time.
In the embodiment illustrated in
In one embodiment, the interfaces between the various encoders, decoder and transcoders conform with Direct Show standards specified by Microsoft Corporation. In other embodiments, the software used to perform the encoding, decoding and transcoding need not comply with such standards.
In the illustrated embodiment, separate processing components are shown for each media source. In other embodiments resources can be shared. For example, a single audio decoder and audio encoder could be used to generate audio chunks from all of the sources. Typically, the entire system can be implemented on a computer using software and connected to a storage device such as a hard disk drive.
In order to utilize the interleaver in the manner described above, the ‘DMNU’ chunk, the ‘DXDT’ chunk, the ‘video’ chunks, the ‘audio’ chunks and the ‘subtitle’ chunks in accordance with embodiments of the present invention must be generated and provided to the interleaver. The process of generating each of the various chunks in a multimedia file in accordance with an embodiment of the present invention is discussed in greater detail below.
The ‘DXDT’ chunk can be generated in any of a number of ways. In one embodiment, ‘meta data’ is entered into data structures via a graphical user interface and then parsed into a ‘DXDT’ chunk. In one embodiment, the ‘meta data’ is expressed as series of subject, predicate, object and authority statements. In another embodiment, the ‘meta data’ statements are expressed in any of a variety of formats. In several embodiments, each ‘meta data’ statement is parsed into a separate chunk. In other embodiments, several ‘meta data’ statements in a first format (such as subject, predicate, object, authority expressions) are parsed into a first chunk and other ‘meta data’ statements in other formats are parsed into separate chunks. In one embodiment, the ‘meta data’ statements are written into an XML configuration file and the XML configuration file is parsed to create the chunks within a ‘DXDT’ chunk.
An embodiment of a system for generating a ‘DXDT’ chunk from a series of ‘meta data’ statements contained within an XML configuration file is shown in
3.3.1. Generating an ‘index’ Chunk
As discussed above, many embodiments of ‘DXDT’ chunks in accordance with the present invention can include an ‘index’ chunk. An ‘index’ chunk can be automatically generated following the completion of the encoding of the audio, video and/or subtitle tracks contained within a multimedia file. In one embodiment, an ‘index’ chunk is generated by referencing encoded video frames and any accompanying audio at approximately evenly spaced intervals within an encoded video track. In other embodiments, video frames associated with scenes and/or chapters within an encoded video track can be identified and used to automatically generate an ‘index’ chunk. In many embodiments, ‘index’ chunks can be generated manually, generated automatically using menu information or generated using any number of algorithms appropriate to the sequence of video indexed by the ‘index’ chunk.
A system that can be used to generate a ‘DMNU’ chunk in accordance with an embodiment of the present invention is illustrated in
The generation of a ‘DMNU’ chunk using the inputs to the menu chunk generating system involves the creation of a number of intermediate files. The media model 422 is used to create an XML configuration file 430 and the media information is used to create a number of AVI files 432. The XML configuration file is created by a model transcoder 434. The AVI files 432 are created by interleaving the video, audio and overlay information using an interleaver 436. The video information is obtained by using a video decoder 438 and a video encoder 440 to decode the video source 424 and recode it in the manner discussed below. The audio information is obtained by using an audio decoder 442 and an audio encoder 444 to decode the audio and encode it in the manner described below. The overlay information is generated using a first transcoder 446 and a second transcoder 448. The first transcoder 446 converts the overlay into a graphical representation such as a standard bitmap and the second transcoder takes the graphical information and formats it as is required for inclusion in the multimedia file. Once the XML file and the AVI files containing the information required to build the menus have been generated, the menu generator 450 can use the information to generate a ‘DMNU’ chunk 358′.
In one embodiment, the media model is an object-oriented model representing all of the menus and their subcomponents. The media model organizes the menus into a hierarchical structure, which allows the menus to be organized by language selection. A media model in accordance with an embodiment of the present invention is illustrated in
The ‘LanguageMenus’ objects organize information for various menus by language selection. All of the ‘Menu’ objects 466 for a given language are associated with the ‘LanguageMenus’ object 463 for that language. Each ‘Menu’ object is associated with a number of ‘Button’ objects 468 and references a number of ‘MediaTrack’ objects 488. The referenced ‘MediaTrack’ objects 488 indicated the background video and background audio for the ‘Menu’ object 466.
Each ‘Button’ object 468 is associated with an ‘Action’ object 470 and a ‘Rectangle’ object 484. The ‘Button’ object 468 also contains a reference to a ‘MediaTrack’ object 488 that indicates the overlay to be used when the button is highlighted on a display. Each ‘Action’ object 470 is associated with a number of objects that can include a ‘MenuTransition’ object 472, a ‘ButtonTransition’ object 474, a ‘ReturnToPlay’ object 476, a ‘Subtitle Selection’ object 478, an ‘AudioSelection’ object 480 and a ‘PlayAction’ object 482. Each of these objects define the response of the menu system to various inputs from a user. The ‘MenuTransition’ object contains a reference to a ‘Menu’ object that indicates a menu that should be transitioned to in response to an action. The ‘ButtonTransition’ object indicates a button that should be highlighted in response to an action. The ‘ReturnToPlay’ object can cause a player to resume playing a feature. The ‘SubtitleSelection’ and ‘AudioSelection’ objects contain references to ‘Title’ objects 487 (discussed below). The ‘PlayAction’ object contains a reference to a ‘Chapter’ object 492 (discussed below). The ‘Rectangle’ object 484 indicates the portion of the screen occupied by the button.
The ‘Media’ object 464 indicates the media information referenced in the menu system. The ‘Media’ object has a ‘MenuTracks’ object 486 and a number of ‘Title’ objects 487 associated with it. The ‘MenuTracks’ object 486 references ‘MediaTrack’ objects 488 that are indicative of the media used to construct the menus (i.e. background audio, background video and overlays).
The ‘Title’ objects 487 are indicative of a multimedia presentation and have a number of ‘Chapter’ objects 492 and ‘MediaSource’ objects 490 associated with them. The ‘Title’ objects also contain a reference to a ‘TranslationLookup’ object 494. The ‘Chapter’ objects are indicative of a certain point in a multimedia presentation and have a number of ‘MediaTrack’ objects 488 associated with them. The ‘Chapter’ objects also contain a reference a ‘TranslationLookup’ object 494. Each ‘MediaTrack’ object associated with a ‘Chapter’ object is indicative of a point in either an audio, video or subtitle track of the multimedia presentation and references a ‘MediaSource’ object 490 and a ‘TransalationLookup’ object 494 (discussed below).
The ‘TranslationTable’ object 465 groups a number of text strings that describe the various parts of multimedia presentations indicated by the ‘Title’ objects, the ‘Chapter’ objects and the ‘MediaTrack’ objects. The ‘TranslationTable’ object 465 has a number of ‘TranslationLookup’ objects 494 associated with it. Each ‘TranslationLookup’ object is indicative of a particular object and has a number of ‘Translation’ objects 496 associated with it. The ‘Translation’ objects are each indicative of a text string that describes the object indicated by the ‘TranslationLookup’ object in a particular language.
A media object model can be constructed using software configured to generate the various objects described above and to establish the required associations and references between the objects.
An XML configuration file is generated from the menu model, which represents all of the menus and their sub-components. The XML configuration file also identifies all the media files used by the menus. The XML can be generated by implementing an appropriate parser application that parses the object model into XML code.
In other embodiments, a video editing application can provide a user with a user interface enabling the direct generation of an XML configuration file without creating a menu model.
In embodiments where another menu system is the basis of the menu model, such as a DVD menu, the menus can be pruned by the user to eliminate menu options relating to content not included in the multimedia file generated in accordance with the practice of the present invention. In one embodiment, this can be done by providing a graphical user interface enabling the elimination of objects from the menu model. In another embodiment, the pruning of menus can be achieved by providing a graphical user interface or a text interface that can edit the XML configuration file.
When the ‘DMNU’ chunk is generated, the media information provided to the menu generator 450 includes the data required to provide the background video, background audio and foreground overlays for the buttons specified in the menu model (see description above). In one embodiment, a video editing application such as VideoWave distributed by Roxio, Inc. of Santa Clara, Calif. is used to provide the source media tracks that represent the video, audio and button selection overlays for each individual menu.
As discussed above, the media tracks that are used as the background video, background audio and foreground button overlays are stored in a single AVI file for one or more menus. The chunks that contain the media tracks in a menu AVI file can be created by using software designed to interleave video, audio and button overlay tracks. The ‘audio’, ‘video’ and ‘overlay’ chunks (i.e. ‘subtitle’ chunks containing overlay information) are interleaved into an AVI format compliant file using an interleaver.
As mentioned above, a separate AVI file can be created for each menu. In other embodiments, other file formats or a single file could be used to contain the media information used to provide the background audio, background video and foreground overlay information.
In one embodiment, a computer is configured to parse information from the XML configuration file to create a ‘WowMenu’ chunk (described above). In addition, the computer can create the ‘MRIF’ chunk (described above) using the AVI files that contain the media for each menu. The computer can then complete the generation of the ‘DMNU’ chunk by creating the necessary references between the ‘WowMenu’ chunk and the media chunks in the ‘MRIF’ chunk. In several embodiments, the menu information can be encrypted. Encryption can be achieved by encrypting the media information contained in the ‘MRIF’ chunk in a similar manner to that described below in relation to ‘video’ chunks. In other embodiments, various alternative encryption techniques are used.
3.3.6. Automatic Generation of Menus from the Object Model
Referring back to
Systems in accordance with several embodiments of the present invention are capable of generating a single XML configuration file containing both ‘meta data’ and menu information and using the XML file to generate the ‘DXDT’ and ‘DMNU’ chunks. These systems derive the XML configuration file using the ‘meta data’ information and the menu object model. In other embodiments, the configuration file need not be in XML.
3.4. Generating ‘audio’ Chunks
The ‘audio’ chunks in the ‘movi’ list chunk of multimedia files in accordance with embodiments of the present invention can be generated by decoding an audio source and then encoding the source into ‘audio’ chunks in accordance with the practice of the present invention. In one embodiment, the ‘audio’ chunks can be encoded using an mp3 codec.
Where the audio source is provided in chunks that don't contain audio information corresponding to the contents of a corresponding ‘video’ chunk, then embodiments of the present invention can re-chunk the audio. A process that can be used to re-chunk audio is illustrated in
3.5. Generating ‘video’ Chunks
As described above the process of creating video chunks can involve decoding the video source and encoding the decoded video into ‘video’ chunks. In one embodiment, each ‘video’ chunk contains information for a single frame of video. The decoding process simply involves taking video in a particular format and decoding the video from that format into a standard video format, which may be uncompressed. The encoding process involves taking the standard video, encoding the video and generating ‘video’ chunks using the encoded video.
A video encoder in accordance with an embodiment of the present invention is conceptually illustrated in
In order to perform motion estimation 506, the system must have knowledge of how the previously processed frame of video will be decoded by a decoding device (e.g. when the compressed video is uncompressed for viewing by a player). This information can be obtained by inverse quantizing 522 the output of the quantizer 512. An inverse DCT 524 can then be performed on the output of the inverse quantizer and the result placed in a frame store 526 for access during the motion estimation process.
Multimedia files in accordance with embodiments of the present invention can also include a number of psychovisual enhancements 528. The psychovisual enhancements can be methods of compressing video based upon human perceptions of vision. These techniques are discussed further below and generally involve modifying the number of bits used by the quantizer to represent various aspects of video. Other aspects of the encoding process can also include psychovisual enhancements.
In one embodiment, the entire encoding system 500 can be implemented using a computer configured to perform the various functions described above. Examples of detailed implementations of these functions are provided below.
The preprocessing operations 502 that are optionally performed by an encoder 500 in accordance with an embodiment of the present invention can use a number of signal processing techniques to improve the quality of the encoded video. In one embodiment, the preprocessing 502 can involve one or all of deinterlacing, temporal/spatial noise reduction and resizing. In embodiments where all three of these preprocessing techniques are used, the deinterlacing is typically performed first followed by the temporal/spatial noise reduction and the resizing.
A video encoder in accordance with an embodiment of the present invention can reduce the number of pixels required to represent a video track by searching for pixels that are repeated in multiple frames. Essentially, each frame in a video typically contains many of the same pixels as the one before it. The encoder can conduct several types of searches for matches in pixels between each frame (as macroblocks, pixels, half-pixels and quarter-pixels) and eliminates these redundancies whenever possible without reducing image quality. Using motion estimation, the encoder can represent most of the picture simply by recording the changes that have occurred since the last frame instead of storing the entire picture for every frame. During motion estimation, the encoder divides the frame it is analyzing into an even grid of blocks, often referred to as ‘Thacroblocks’. For each ‘Thacroblock’ in the frame, the encoder can try to find a matching block in the previous frame. The process of trying to find matching blocks is called a ‘motion search’. The motion of the ‘Thacroblock’ can be represented as a two dimensional vector, i.e. an (x,y) representation. The motion search algorithm can be performed with various degrees of accuracy. A whole-pel search is one where the encoder will try to locate matching blocks by stepping through the reference frame in either dimension one pixel at a time. Ina half-pixel search, the encoder searches for a matching block by stepping through the reference frame in either dimension by half of a pixel at a time. The encoder can use quarter-pixels, other pixel fractions or searches involving a granularity of greater than a pixel.
The encoder embodiment illustrated in
Motion compensation is performed by taking the blocks and vectors generated as a result of motion estimation. The result is an approximation of the encoded image that can be matched to the actual image by providing additional texture information.
The DCT and inverse DCT performed by the encoder illustrated in
The DCT is a method of transforming a set of spatial-domain data points to a frequency domain representation. In the case of video compression, a 2-dimensional DCT converts image blocks into a form where redundancies are more readily exploitable. A frequency domain block can be a sparse matrix that is easily compressed by entropy coding.
The DCT coefficients can be modified to improve the quality of the quantized image by reducing quantization noise in areas where it is readily apparent to a human viewer. In addition, file size can be reduced by increasing quantization noise in portions of the image where it is not readily discernable by a human viewer.
Encoders in accordance with an embodiment of the present invention can perform what is referred to as a ‘slow’ psychovisual enhancement. The ‘slow’ psychovisual enhancement analyzes blocks of the video image and decides whether allowing some noise there can save some bits without degrading the video's appearance. The process uses one metric per block. The process is referred to as a ‘slow’ process, because it performs a considerable amount of computation to avoid blocking or ringing artifacts.
Other embodiments of encoders in accordance with embodiments of the present invention implement a ‘fast’ psychovisual enhancement. The ‘fast’ psychovisual enhancement is capable of controlling where noise appears within a block and can shape quantization noise.
Both the ‘slow’ and ‘fast’ psychovisual enhancements are discussed in greater detail below. Other psychovisual enhancements can be performed in accordance with embodiments of the present invention including enhancements that control noise at image edges and that seek to concentrate higher levels of quantization noise in areas of the image where it is not readily apparent to human vision.
The ‘slow’ psychovisual enhancement analyzes blocks of the video image and determines whether allowing some noise can save bits without degrading the video's appearance. In one embodiment, the algorithm includes two stages. The first involves generation of a differentiated image for the input luminance pixels. The differentiated image is generated in the manner described below. The second stage involves modifying the DCT coefficients prior to quantization.
Each pixel p′xy of the differentiated image is computed from the uncompressed source pixels, pxy, according to the following:
p′
xy=max(|px+1y−pxy|,|px−1y−pxy|,|pxy+1−pxy|,|pxy−1−pxy)|
where
p′xy will be in the range 0 to 255 (assuming 8 bit video).
The modification of the DCT coefficients can involve computation of a block ringing factor, computation of block energy and the actual modification of the coefficient values.
For each block of the image, a “ringing factor” is calculated based on the local region of the differentiated image. In embodiments where the block is defined as an 8×8 block, the ringing factor can be determined using the following method.
Initially, a threshold is determined based on the maximum and minimum luminance pixels values within the 8×8 block:
thresholdblock=floor((maxblock−minblock)/8)+2
The differentiated image and the threshold are used to generate a map of the “flat” pixels in the block's neighborhood. The potential for each block to have a different threshold prevents the creation of a map of flat pixels for the entire frame. The map is generated as follows:
flatxy=1 when p′xy<thresholdblock
flatxy=0 otherwise
The map of flat pixels is filtered according to a simple logical operation:
flat′xy=1 when flatxy=1 and flatx−1y=1 and flatxy−1=1 and flatx−1y−1=1 flat′xy, otherwise
The flat pixels in the filtered map are then counted over the 9×9 region that covers the 8×8 block.
flatcountblock=Σflat′xy for 0=x=8 and 0=y=8
The risk of visible ringing artifacts can be evaluated using the following expression:
ringingbriskblock=((flatcountblock−10)×256+20)/40
The 8×8 block's ringing factor can then be derived using the following expression:
The energy for blocks of the image can be calculated using the following procedure. In several embodiments, 8×8 blocks of the image are used.
A forward DCT is performed on the source image:
T=fDCT(S)
where S is the 64 source-image luminance values of the 8×8 block in question and T is the transformed version of the same portion of the source image.
The energy at a particular coefficient position is defined as the square of that coefficient's value:
e
k
=t
k
2 for 0=k=63
where tk is the kth coefficient of transformed block T.
The modification of the DCT coefficients can be performed in accordance with the following process. In several embodiments, the process is performed for every non-zero AC DCT coefficient before quantization. The magnitude of each coefficient is changed by a small delta, the value of the delta being determined according to psychovisual techniques.
The DCT coefficient modification of each non-zero AC coefficient ck is performed by calculating an energy based on local and block energies using the following formula:
energyk=max(ak×ek,0.12×totalenergy)
where ak is a constant whose value depends on the coefficient position as described in the following table:
The energy can be modified according to the block's ringing factor using the following relationship:
energy′k=ringingfactor×energyk
The resulting value is shifted and clipped before being used as an input to a look-up table (LUT).
e
k=min(1023,4×energy′k)
d
k=LUTi where i=ek
The look-up table is computed as follows:
LUTi=min(floor(ktexture×((i+0.5)/4)1/2+kflat×offset),2×QP)
The value ‘offset’ depends on quantizer, Qp, as described in the following table:
The variable ktexture and kflat control the strength of the of the psychovisual effect in flat and textured regions respectively. In one embodiment, they take values in the range 0 to 1, with 0 signifying no effect and 1 meaning full effect. In one embodiment, the values for ktexture and kflat are established as follows:
Luminance:
k
texture=1.0
k
flat=1.0
Chrominance:
k
texture=1.0
k
flat=0.0
The output from the look-up table (dk) is used to modify the magnitude of the DCT coefficient by an additive process:
c′
k
=c
k−min(dk,|ck|)×sgn(ck)
Finally, the DCT coefficient ck is substituted by the modified coefficient c′k and passed onwards for quantization.
A ‘fast’ psychovisual enhancement can be performed on the DCT coefficients by computing an ‘importance’ map for the input luminance pixels and then modifying the DCT coefficients.
An ‘importance’ map can be generated by calculating an ‘importance’ value for each pixel in the luminance place of the input video frame. In several embodiments, the ‘importance’ value approximates the sensitivity of the human eye to any distortion located at that particular pixel. The ‘importance’ map is an array of pixel ‘importance’ values.
The ‘importance’ of a pixel can be determined by first calculating the dynamic range of a block of pixels surrounding the pixel (dxy). In several embodiments the dynamic range of a 3×3 block of pixels centered on the pixel location (x, y) is computed by subtracting the value of the darkest pixel in the area from the value of the lightest pixel in the area.
The ‘importance’ of a pixel (mxy) can be derived from the pixel's dynamic range as follows:
m
Xy=0.08/max(dxy,3)+0.001
In one embodiment, the modification of the DCT coefficients involves the generation of basis-function energy matrices and delta look up tables.
A set of basis-function energy matrices can be used in modifying the DCT coefficients. These matrices contain constant values that may be computed prior to encoding. An 8×8 matrix is used for each of the 64 DCT basis functions. Each matrix describes how every pixel in an 8×8 block will be impacted by modification of its corresponding coefficient. The kth basis-function energy matrix is derived by taking an 8×8 matrix Ak with the corresponding coefficient set to 100 and the other coefficients set to 0.
where
n represents the coefficient position within the 8×8 matrix; 0=n=63
An inverse DCT is performed on the matrix to yield a further 8×8 matrix A′k. The elements of the matrix (a′kn) represent the kth DCT basis function.
A′
k
=iDCT(Ak)
Each value in the transformed matrix is then squared:
b
kn
=a′
kn
2 for 0=n=63
The process is carried out 64 times to produce the basis function energy matrices Bk, 0=k=63, each comprising 64 natural values. Each matrix value is a measure of how much a pixel at the nth position in the 8×8 block will be impacted by any error or modification of the coefficient k.
A look-up table (LUT) can be used to expedite the computation of the coefficient modification delta. The contents of the table can be generated in a manner that is dependent upon the desired strength of the ‘fast’ psychovisual enhancement and the quantizer parameter (Qp).
The values of the look-up table can be generated according to the following relationship:
LUTi=min(floor(128×ktexture×strength/(i+0.5)+kflat×offset+0.5),2×Qp)
where
i is the position within the table, 0=i=1023.
strength and offset depend on the quantizer, Qp, as described in the following table:
The variable ktexture and kflat control the strength of the of the psychovisual effect in flat and textured regions respectively. In one embodiment, they take values in the range 0 to 1, with 0 signifying no effect and 1 meaning full effect. In one embodiment, the values for ktexture and kflat are established as follows:
Luminance:
k
texture=1.0
k
flat=1.0
Chrominance:
k
texture=1.0
k
flat=0.0
The DCT coefficients can be modified using the values calculated above. In one embodiment, each non-zero AC DCT coefficient is modified in accordance with the following procedure prior to quantization.
Initially, an ‘energy’ value (ek) is computed by taking the dot product of the corresponding basis function energy matrix and the appropriate 8×8 block from the importance map. This ‘energy’ is a measure of how quantization errors at the particular coefficient would be perceived by a human viewer. It is the sum of the product of pixel importance and pixel basis-function energy:
e
k
=M·B
k
where
M contains the 8×8 block's importance map values; and
Bk is the kth basis function energy matrix.
The resulting ‘energy’ value is shifted and clipped before being used as an index (dk) into the delta look-up table.
e′
k=min[1023,floor(ek/32768)]
d
k=LUTi
where
i=e′k
The output of the delta look-up table is used to modify the magnitude of the DCT coefficient by an additive process:
c′
k
=c
k−min(dk,|ck|)×sign(ck)
The DCT coefficient ck is substituted with the modified c′k and passed onwards for quantization.
Encoders in accordance with embodiments of the present invention can use a standard quantizer such as a the quantizer defined by the International Telecommunication Union as Video Coding for Low Bitrate Communication, ITU-T Recommendation H.263, 1996.
Some encoders in accordance with embodiments of the present invention, use a psychovisual enhancement that exploits the psychological effects of human vision to achieve more efficient compression. The psychovisual effect can be applied at a frame level and a macroblock level.
When applied at a frame level, the enhancement is part of the rate control algorithm and its goal is to adjust the encoding so that a given amount of bit rate is best used to ensure the maximum visual quality as perceived by human eyes. The frame rate psychovisual enhancement is motivated by the theory that human vision tends to ignore the details when the action is high and that human vision tends to notice detail when an image is static. In one embodiment, the amount of motion is determined by looking at the sum of absolute difference (SAD) for a frame. In one embodiment, the SAD value is determined by summing the absolute differences of collocated luminance pixels of two blocks. In several embodiments, the absolute differences of 16×16 pixel blocks is used. In embodiments that deal with fractional pixel offsets, interpolation is performed as specified in the MPEG-4 standard (an ISO/IEC standard developed by the Moving Picture Experts Group of the ISO/IEC), before the sum of absolute differences is calculated.
The frame-level psychovisual enhancement applies only to the P frames of the video track and is based on SAD value of the frame. During the encoding, the psychovisual module keeps a record of the average SAD (i.e.
For each P frame of the video track encoded, the frame quantizer Q (obtained from the rate control module) will have a psychovisual correction applied to it. In one embodiment, the process involves calculating a ratio R using the following formula:
where
I is a constant and is currently set to 0.5. The R is clipped to within the bound of [−1, 1].
The quantizer is then adjusted according to the ration R, via the calculation shown below:
Q
adj
=Q┌Q·(1+R·Sframe)┘
where
Sframe is a strength constant for the frame level psychovisual enhancements.
The Sframe constant determines how strong an adjustment can be for the frame level psychovisual. In one embodiment of the codec, the option of setting Sframe to 0.2, 0.3 or 0.4 is available.
Encoders in accordance with embodiments of the present invention that utilize a psychovisual enhancement at the macroblock level attempt to identify the macroblocks that are prominent to the visual quality of the video for a human viewer and attempt to code those macroblocks with higher quality. The effect of the macroblock level psychovisual enhancements it to take bits away from the less important parts of a frame and apply them to more important parts of the frame. In several embodiments, enhancements are achieved using three technologies, which are based on smoothness, brightness and the macroblock SAD. In other embodiments any of the techniques alone or in combination with another of the techniques or another technique entirely can be used.
In one embodiment, all three of the macroblock level psychovisual enhancements described above share a common parameter, SMB, which controls the strength of the macroblock level psychovisual enhancement. The maximum and minimum quantizer for the macroblocks are then derived from the strength parameter and the frame quantizer Qframe via the calculations shown below:
where
QMBMax is the maximum quantizer
QMBMax is the minimum quantizer
The values QMBMax and QMBMax define the upper and lower bounds to the macroblock quantizers for the entire frame. In one embodiment, the option of setting the value SMB to any of the values 0.2, 0.3 and 0.4 is provided. In other embodiments, other values for SMB can be utilized.
In embodiments where psychovisual enhancement is performed based on the brightness of the macroblocks, the encoder attempts to encode brighter macroblocks with greater quality. The theoretical basis of this enhancement is that relatively dark parts of the frame are more or less ignored by human viewers. This macroblock psychovisual enhancement is applied to I frames and P frames of the video track. For each frame, the encoder looks through the whole frame first. The average brightness (
T
BRLower
=
T
BRUpper
=
The brightness enhancement is then applied based on the two thresholds using the conditions stated below to generate an intended quantizer (QMB) for the macroblock:
Q
MB
=Q
MBMin when BR>TBRUpper
Q
MB
=Q
frame when TBRLower≤BR≤TBRUpper and
Q
MB
=Q
MBMax when BR<TBRLower
where
BR is the brightness value for that particular macroblock
In embodiments where the encoder is compliant with the MPEG-4 standard, the macroblock level psychovisual brightness enhancement technique cannot change the quantizer by more than ±2 from one macroblock to the next one. Therefore, the calculated QMB may require modification based upon the quantizer used in the previous macroblock.
Encoders in accordance with embodiments of the present invention that include a smoothness psychovisual enhancement, modify the quantizer based on the spatial variation of the image being encoded. Use of a smoothness psychovisual enhancement can be motivated by the theory that human vision has an increased sensitivity to quantization artifacts in smooth parts of an image. Smoothness psychovisual enhancement can, therefore, involve increasing the number of bits to represent smoother portions of the image and decreasing the number of bits where there is a high degree of spatial variation in the image.
In one embodiment, the smoothness of a portion of an image is measured as the average difference in the luminance of pixels in a macroblock to the brightness of the macroblock (
The following smoothness enhancement is performed (546) based on the threshold.
Q
MB
=Q
frame when DR≥TDR, and
Q
MB
=Q
MBMin when DR<TDR
where
QMB is the intended quantizer for the macroblock
DR is the deviation value for the macroblock(i.e. mean luminance−mean brightness)
Embodiments that encode files in accordance with the MPEG-4 standard are limited as described above in that the macroblock level quantizer change can be at most ±2 from one macroblock to the next.
Encoders in accordance with embodiments of the present invention can utilize a macroblock SAD psychovisual enhancement. A macroblock SAD psychovisual enhancement can be used to increase the detail for static macroblocks and allow decreased detail in portions of a frame that are used in a high action scene.
A process for performing a macroblock SAD psychovisual enhancement in accordance with an embodiment of the present invention is illustrated in
T
MBSADLower=
T
MBSADUpper=
where
TMBSADLower is the lower threshold
TMBSADUpper is the upper threshold, which may be bounded by 1024 if necessary
The macroblock SAD enhancement is then applied (576) based on these two thresholds according to the following conditions:
Q
MB
=Q
MBMax when
Q
MB
=Q
frame when TMADLower≤MBSAD≤TMBSADUpper
Q
MB
=Q
MBMin when MBSAD<TMBSADLower
where
QMB is the intended quantizer for the macroblock
MBSAD is the SAD value for that particular macroblock
Embodiments that encode files in accordance with the MPEG-4 specification are limited as described above in that the macroblock level quantizer change can be at most ±2 from one macroblock to the next.
The rate control technique used by an encoder in accordance with an embodiment of the present invention can determine how the encoder uses the allocated bit rate to encode a video sequence. An encoder will typically seek to encode to a predetermined bit rate and the rate control technique is responsible for matching the bit rate generated by the encoder as closely as possible to the predetermined bit rate. The rate control technique can also seek to allocate the bit rate in a manner that will ensure the highest visual quality of the video sequence when it is decoded. Much of rate control is performed by adjusting the quantizer. The quantizer determines how finely the encoder codes the video sequence. A smaller quantizer will result in higher quality and higher bit consumption. Therefore, the rate control algorithm seeks to modify the quantizer in a manner that balances the competing interests of video quality and bit consumption.
Encoders in accordance with embodiments of the present invention can utilize any of a variety of different rate control techniques. In one embodiment, a single pass rate control technique is used. In other embodiments a dual (or multiple) pass rate control technique is used. In addition, a ‘video buffer verified’ rate control can be performed as required. Specific examples of these techniques are discussed below. However, any rate control technique can be used in an encoder in accordance with the practice of the present inventions.
An embodiment of a one pass rate control technique in accordance with an embodiment of the present invention seeks to allow high bit rate peaks for high motion scenes. In several embodiments, the one pass rate control technique seeks to increase the bit rate slowly in response to an increase in the amount of motion in a scene and to rapidly decrease the bit rate in response to a reduction in the motion in a scene.
In one embodiment, the one pass rate control algorithm uses two averaging periods to track the bit rate. A long-term average to ensure overall bit rate convergence and a short-term average to enable response to variations in the amount of action in a scene.
A one pass rate control technique in accordance with an embodiment of the present invention is illustrated in
As discussed above, the encoder is initialized (584) with a variety of parameters. These parameters are the ‘bit rate’, the ‘frame rate’, the ‘Max Key Frame Interval’, the ‘Maximum Quantizer’, the ‘Minimum Quantizer’, the ‘averaging period’, the ‘reaction period’ and the ‘down/up ratio’. The following is a discussion of each of these parameters.
The ‘bit rate’ parameter sets the target bit rate of the encoding.
The ‘frame rate’ defines the period between frames of video.
The ‘Max Key Frame Interval’ specifies the maximum interval between the key frames. The key frames are normally automatically inserted in the encoded video when the codec detects a scene change. In circumstances where a scene continues for a long interval without a single cut, key frames can be inserted in insure that the interval between key frames is always less or equal to the ‘Max Key Frame Interval’. In one embodiment, the ‘Max Key Frame Interval’ parameter can be set to a value of 300 frames. In other embodiments, other values can be used.
The ‘Maximum Quantizer’ and the ‘Minimum Quantizer’ parameters set the upper and lower bound of the quantizer used in the encoding. In one embodiment, the quantizer bounds are set at values between 1 and 31.
The ‘averaging period’ parameter controls the amount of video that is considered when modifying the quantizer. A longer averaging period will typically result in the encoded video having a more accurate overall rate. In one embodiment, an ‘averaging period’ of 2000 is used. Although in other embodiments other values can be used.
3.5.5.1.6. The ‘reaction Period’
The ‘reaction period’ parameter determines how fast the encoder adapts to changes in the motion in recent scenes. A longer ‘reaction period’ value can result in better quality high motion scenes and worse quality low motion scenes. In one embodiment, a ‘reaction period’ of 10 is used. Although in other embodiments other values can be used.
The ‘down/up ratio’ parameter controls the relative sensitivity for the quantizer adjustment in reaction to the high or low motion scenes. A larger value typically results in higher quality high motion scenes and increased bit consumption. In one embodiment, a ‘down/up ratio’ of 20 is used. Although in other embodiments, other values can be used.
As discussed above, the one pass rate control technique involves the calculation of a quantizer value after the encoding of each frame. The following is a description of a technique in accordance with an embodiment of the present invention that can be used to update the quantizer value.
The encoder maintains two exponential moving averages having periods equal to the ‘averaging period’ (Paverage) and the ‘reaction period’ (Preaction) a moving average of the bit rate. The two exponential moving averages can be calculated according to the relationship:
where
At is the average at instance t;
At-1 is the average at instance t-T (usually the average in the previous frame);
T represents the interval period (usually the frame time); and
P is the average period, which can be either Paverage and or Preaction.
The above calculated moving average is then adjusted into bit rate by dividing by the time interval between the current instance and the last instance in the video, using the following calculation:
where
Rt is the bitrate;
At is either of the moving averages; and
T is the time interval between the current instance and last instance (it is usually the inverse of the frame rate).
The encoder can calculate the target bit rate (Rtarget) of the next frame as follows:
R
target
=R
overall+(Roverall−Raverage)
where
Roverall is the overall bit rate set for the whole video; and
Raverage is the average bit rate using the long averaging period.
In several embodiments, the target bit rate is lower bounded by 75% of the overall bit rate. If the target bit rate drops below that bound, then it will be forced up to the bound to ensure the quality of the video.
The encoder then updates the internal quantizer based on the difference between Rtarget and Rreaction. If Rreaction is less than Rtarget, then there is a likelihood that the previous frame was of relatively low complexity. Therefore, the quantizer can be decreased by performing the following calculation:
When Rreaction is greater than Rtarget, there is a significant likelihood that previous frame possessed a relatively high level of complexity. Therefore, the quantizer can be increased by performing the following calculation:
where
S is the ‘up/down ratio’.
The algorithm described above can also be applied to B-VOP encoding. When B-VOP is enabled in the encoding, the quantizer for the B-VOP (QB) is chosen based on the quantizer of the P-VOP (QP) following the B-VOP. The value can be obtained in accordance with the following relationships:
Q
B=2·QP for QP≤4
Q
B=5+¾·QP for 4<QP≤20
Q
B
=Q
P for QP≥20
Encoders in accordance with an embodiment of the present invention that use a two (or multiple) pass rate control technique can determine the properties of a video sequence in a first pass and then encode the video sequence with knowledge of the properties of the entire sequence. Therefore, the encoder can adjust the quantization level for each frame based upon its relative complexity compared to other frames in the video sequence.
A two pass rate control technique in accordance with an embodiment of the present invention, the encoder performs a first pass in which the video is encoded in accordance with the one pass rate control technique described above and the complexity of each frame is recorded (any of a variety of different metrics for measuring complexity can be used). The average complexity and, therefore, the average quantizer (Qref) can be determined based on the first. In the second pass, the bit stream is encoded with quantizers determined based on the complexity values calculated during the first pass.
The quantizer Q for I-VOPs is set to 0.75×Qref, provided the next frame is not an I-VOP. If the next frame is also an I-VOP, the Q(for the current frame) is set to 1.25×Qref.
The quantizer for the P-VOPs can be determined using the following expression.
where
Ccomplexity is the complexity of the frame;
F(x) is a function that provides the number which the complexity of the frame must be multiplied to give the number of bits required to encode the frame using a quantizer with a quantization value x;
F−1(x) is the inverse function of F(x); and
k is the strength parameter.
The following table defines an embodiment of a function F(Q) that can be used to generator the factor that the complexity of a frame must be multiplied by in order to determine the number of bits required to encode the frame using an encoder with a quantizer Q.
If the strength parameter k is chosen to be 0, then the result is a constant quantizer. When the strength parameter is chosen to be 1, the quantizer is proportional to Ccomplexity. Several encoders in accordance with embodiments of the present invention have a strength parameter k equal to 0.5.
The quantizer Q for the B-VOPs can be chosen using the same technique for choosing the quantizer for B-VOPs in the one pass technique described above.
The number of bits required to represent a frame can vary depending on the characteristics of the video sequence. Most communication systems operate at a constant bit rate. A problem that can be encountered with variable bit rate communications is allocating sufficient resources to handle peaks in resource usage. Several encoders in accordance with embodiments of the present invention encode video with a view to preventing overflow of a decoder video buffer, when the bit rate of the variable bit rate communication spikes.
The objectives of video buffer verifier (VBV) rate control can include generating video that will not exceed a decoder's buffer when transmitted. In addition, it can be desirable that the encoded video match a target bit rate and that the rate control produces high quality video.
Encoders in accordance with several embodiments of the present invention provide a choice of at least two VBV rate control techniques. One of the VBV rate control techniques is referred to as causal rate control and the other technique is referred to as Nth pass rate control.
Causal VBV rate control can be used in conjunction with a one pass rate control technique and generates outputs simply based on the current and previous quantizer values.
An encoder in accordance with an embodiment of the present invention includes causal rate control involving setting the quantizer for frame n (i.e. Qn) according to the following relationship.
where
Q′n is the quantizer estimated by the single pass rate control;
Xbitrate is calculated by determining a target bit rate based on the drift from the desired bit rate;
Xvelocity is calculated based on the estimated time until the VBV buffer over- or under-flows;
Xsize is applied on the result of P-VOPs only and is calculated based on the rate at which the size of compressed P-VOPs is changing over time;
Xdrift is the drift from the desired bit rate.
In several embodiments, the causal VBV rate control may be forced to drop frames and insert stuffing to respect the VBV model. If a compressed frame unexpectedly contains too many or two few bits, then it can be dropped or stuffed.
Nth pass VBV rate control can be used in conjunction with a multiple pass rate control technique and it uses information garnered during previous analysis of the video sequence. Encoders in accordance with several embodiments of the present invention perform Nth pass VBV rate control according to the process illustrated in
In one embodiment, the first pass uses some form of causal rate control and data is recorded for each frame concerning such things as the duration of the frame, the coding type of the frame, the quantizer used, the motion bits produced and the texture bits produced. In addition, global information such as the timescale, resolution and codec settings can also be recorded.
Information from the analysis is used to generate a map of the video sequence. The map can specify the coding type used for each frame (I/B/P) and can include data for each frame concerning the duration of the frame, the motion complexity and the texture complexity. In other embodiments, the map may also contain information enabling better prediction of the influence of quantizer and other parameters on compressed frame size and perceptual distortion. In several embodiments, map generation is performed after the N−1th pass is completed.
The map can be used to plan a strategy as to how the Nth pass rate control will operate. The ideal level of the VBV buffer after every frame is encoded can be planned. In one embodiment, the strategy generation results in information for each frame including the desired compressed frame size, an estimated frame quantizer. In several embodiments, strategy generation is performed after map generation and prior to the Nth pass.
In one embodiment, the strategy generation process involves use of an iterative process to simulate the encoder and determine desired quantizer values for each frame by trying to keep the quantizer as close as possible to the median quantizer value. A binary search can be used to generate a base quantizer for the whole video sequence. The base quantizer is the constant value that causes the simulator to achieve the desired target bit rate. Once the base quantizer is found, the strategy generation process involves consideration of the VBV constrains. In one embodiment, a constant quantizer is used if this will not modify the VBV constrains. In other embodiments, the quantizer is modulated based on the complexity of motion in the video frames. This can be further extended to incorporate masking from scene changes and other temporal effects.
In one embodiment, the in-loop Nth pass rate control uses the strategy and uses the map to make the best possible prediction of the influence of quantizer and other parameters on compressed frame size and perceptual distortion. There can be a limited discretion to deviate from the strategy to take short-term corrective strategy. Typically, following the strategy will prevent violation of the VBV model. In one embodiment, the in-loop Nth pass rate control uses a PID control loop. The feedback in the control loop is the accumulated drift from the ideal bitrate.
Although the strategy generation does not involve dropping frames, the in-loop Nth rate control may drop frames if the VBV buffer would otherwise underflow. Likewise, the in-loop Nth pass rate control can request video stuffing to be inserted to prevent VBV overflow.
In one embodiment, AD/DC prediction is performed in a manner that is compliant with the standard referred to as ISO/IEC 14496-2:2001(E), section 7.4.3. (DC and AC prediction) and 7.7.1. (field DC and AC prediction).
An encoder in accordance with an embodiment of the present invention can perform texture coding in a manner that is compliant with the standard referred to as ISO/IEC 14496-2:2001(E), annex B (variable length codes) and 7.4.1. (variable length decoding).
An encoder in accordance with an embodiment of the present invention can perform motion coding in a manner that is compliant with the standard referred to as ISO/IEC 14496-2:2001(E), annex B (variable length codes) and 7.6.3. (motion vector decoding).
The video track can be considered a sequence of frames 1 to N. Systems in accordance with embodiments of the present invention are capable of encoding the sequence to generate a compressed bitstream. The bitstream is formatted by segmenting it into chunks 1 to N. Each video frame n has a corresponding chunk n.
The chunks are generated by appending bits from the bitstream to chunk n until it, together with the chunks 1 through n−1 contain sufficient information for a decoder in accordance with an embodiment of the present invention to decode the video frame n. In instances where sufficient information is contained in chunks 1 through n−1 to generate video frame n, an encoder in accordance with embodiments of the present invention can include a marker chunk. In one embodiment, the marker chunk is a not-coded P-frame with identical timing information as the previous frame.
An encoder in accordance with an embodiment of the present invention can take subtitles in one of a series of standard formats and then converts the subtitles to bit maps. The information in the bit maps is then compressed using run length encoding. The run length encoded bit maps are the formatted into a chunk, which also includes information concerning the start time and the stop time for the particular subtitle contained within the chunk. In several embodiments, information concerning the color, size and position of the subtitle on the screen can also be included in the chunk. Chunks can be included into the subtitle track that set the palette for the subtitles and that indicate that the palette has changed. Any application capable of generating a subtitle in a standard subtitle format can be used to generate the text of the subtitles. Alternatively, software can be used to convert text entered by a user directly into subtitle information.
Once the interleaver has received all of the chunks described above, the interleaver builds a multimedia file. Building the multimedia file can involve creating a ‘CSET’ chunk, an ‘INFO’ list chunk, a ‘hdrl’ chunk, a ‘movi’ list chunk and an idx1 chunk. Methods in accordance with embodiments of the present invention for creating these chunks and for generating multimedia files are described below.
As described above, the ‘CSET’ chunk is optional and can generated by the interleaver in accordance with the AVI Container Format Specification.
As described above, the ‘INFO’ list chunk is optional and can be generated by the interleaver in accordance with the AVI Container Format Specification.
3.7.3. Generating the ‘hdrl’ List Chunk
The ‘hdrl’ list chunk is generated by the interleaver based on the information in the various chunks provided to the interleaver. The ‘hdrl’ list chunk references the location within the file of the referenced chunks. In one embodiment, the ‘hdrl’ list chunk uses file offsets in order to establish references.
3.7.4. Generating the ‘movi’ List Chunk
As described above, ‘movi’ list chunk is created by encoding audio, video and subtitle tracks to create ‘audio’, ‘video’ and ‘subtitle chunks and then interleaving these chunks. In several embodiments, the ‘movi’ list chunk can also include digital rights management information.
A variety of rules can be used to interleave the audio, video and subtitle chunks. Typically, the interleaver establishes a number of queues for each of the video and audio tracks. The interleaver determines which queue should be written to the output file. The queue selection can be based on the interleave period by writing from the queue that has the lowest number of interleave periods written. The interleaver may have to wait for an entire interleave period to be present in the queue before the chunk can be written to the file.
In one embodiment, the generated ‘audio,’ ‘video’ and ‘subtitle’ chunks are interleaved so that the ‘audio’ and ‘subtitle’ chunks are located within the file prior to the ‘video’ chunks containing information concerning the video frames to which they correspond. In other embodiments, the ‘audio’ and ‘subtitle’ chunks can be located after the ‘video’ chunks to which they correspond. The time differences between the location of the ‘audio,’ ‘video’ and ‘subtitle’ chunks is largely dependent upon the buffering capabilities of players that are used to play the devices. In embodiments where buffering is limited or unknown, the interleaver interleaves the ‘audio,’ ‘video’ and ‘subtitle’ chunks such that the ‘audio’ and ‘subtitle’ chunks are located between ‘video’ chunks, where the ‘video’ chunk immediately following the ‘audio’ and ‘subtitle’ chunk contains the first video frame corresponding to the audio or subtitle.
In embodiments where DRM is used to protect the video content of a multimedia file, the DRM information can be generated concurrently with the encoding of the video chunks. As each chunk is generated, the chunk can be encrypted and a DRM chunk generated containing information concerning the encryption of the video chunk.
An interleaver in accordance with an embodiment of the present invention interleaves a DRM chunk containing information concerning the encryption of a video chunk prior to the video chunk. In one embodiment, the DRM chunk for video chunk n is located between video chunk n−1 and video chunk n. In other embodiments, the spacing of the DRM before and after the video chunk n is dependent upon the amount of buffering provided within device decoding the multimedia file.
3.7.5. Generating the ‘idx1’ Chunk
Once the ‘movi’ list chunk has been generated, the generation of the ‘idx1’ chunk is a simple process. The ‘idx1’ chunk is created by reading the location within the ‘movi’ list chunk of each ‘data’ chunk. This information is combined with information read from the ‘data’ chunk concerning the track to which the ‘data’ chunk belongs and the content of the ‘data’ chunk. All of this information is then inserted into the ‘idx1’ chunk in a manner appropriate to whichever of the formats described above is being used to represent the information.
Once a multimedia file is generated, the file can be distributed over any of a variety of networks. The fact that in many embodiments the elements required to generate a multimedia presentation and menus, amongst other things, are contained within a single file simplifies transfer of the information. In several embodiments, the multimedia file can be distributed separately from the information required to decrypt the contents of the multimedia file.
In one embodiment, multimedia content is provided to a first server and encoded to create a multimedia file in accordance with the present invention. The multimedia file can then be located either at the first server or at a second server. In other embodiments, DRM information can be located at the first server, the second server or a third server. In one embodiment, the first server can be queried to ascertain the location of the encoded multimedia file and/or to ascertain the location of the DRM information.
Information from a multimedia file in accordance with an embodiment of the present invention can be accessed by a computer configured using appropriate software, a dedicated player that is hardwired to access information from the multimedia file or any other device capable of parsing an AVI file. In several embodiments, devices can access all of the information in the multimedia file. In other embodiments, a device may be incapable of accessing all of the information in a multimedia file in accordance with an embodiment of the present invention. In a particular embodiment, a device is not capable of accessing any of the information described above that is stored in chunks that are not specified in the AVI file format. In embodiments where not all of the information can be accessed, the device will typically discard those chunks that are not recognized by the device.
Typically, a device that is capable of accessing the information contained in a multimedia file in accordance with an embodiment of the present invention is capable of performing a number of functions. The device can display a multimedia presentation involving display of video on a visual display, generate audio from one of potentially a number of audio tracks on an audio system and display subtitles from potentially one of a number of subtitle tracks. Several embodiments can also display menus on a visual display while playing accompanying audio and/or video. These display menus are interactive, with features such as selectable buttons, pull down menus and sub-menus. In some embodiments, menu items can point to audio/video content outside the multimedia file presently being accessed. The outside content may be either located local to the device accessing the multimedia file or it may be located remotely, such as over a local area, wide are or public network. Many embodiments can also search one or more multimedia files according to ‘meta data’ included within the multimedia file(s) or ‘meta data’ referenced by one or more of the multimedia files.
Given the ability of multimedia files in accordance with embodiments of the present invention to support multiple audio tracks, multiple video tracks and multiple subtitle tracks, the display of a multimedia presentation using such a multimedia file that combines video, audio and/or subtitles can require selection of a particular audio track, video track and/or subtitle track either through a visual menu system or a pull down menu system (the operation of which are discussed below) or via the default settings of the device used to generate the multimedia presentation. Once an audio track, video track and potentially a subtitle track are selected, the display of the multimedia presentation can proceed.
A process for locating the required multimedia information from a multimedia file including DRM and displaying the multimedia information in accordance with an embodiment of the present invention is illustrated in
In several embodiments the chosen audio track can include multiple channels to provide stereo or surround sound audio. When a subtitle track is chosen to be displayed, a determination can be made as to whether the previous video frame included a subtitle (this determination may be made in any of a variety of ways that achieves the outcome of identifying a previous ‘subtitle’ chunk that contained subtitle information that should be displayed over the currently decoded video frame). If the previous subtitle included a subtitle and the timing information for the subtitle indicates that the subtitle should be displayed with the current frame, then the subtitle is superimposed on the decoded video frame. If the previous frame did not include a subtitle or the timing information for the subtitle on the previous frame indicates that the subtitle should not be displayed in conjunction with the currently decoded frame, then a ‘subtitle’ chunk for the selected subtitle track is sought. If a ‘subtitle’ chunk is located, then the subtitle is superimposed on the decoded video. The video (including any superimposed subtitles) is then displayed with the accompanying audio.
Returning to the discussion of
In several embodiments, a device can seek to a particular portion of the multimedia information (e.g. a particular scene of a movie with a particular accompanying audio track and optionally a particular accompanying subtitle track) using information contained within the ‘hdrl’ chunk of a multimedia file in accordance with the present invention. In many embodiments, the decoding of the ‘video’ chunk, ‘audio’ chunk and/or ‘subtitle’ chunk can be performed in parallel with other tasks.
An example of a device capable of accessing information from the multimedia file and displaying video in conjunction with a particular audio track and/or a particular subtitle track is a computer configured in the manner described above using software. Another example is a DVD player equipped with a codec that includes these capabilities. In other embodiments, any device configured to locate or select (whether intentionally or arbitrarily) ‘data’ chunks corresponding to particular media tracks and decode those tracks for presentation is capable of generating a multimedia presentation using a multimedia file in accordance with the practice of the present invention.
In several embodiments, a device can play multimedia information from a multimedia file in combination with multimedia information from an external file. Typically, such a device would do so by sourcing an audio track or subtitle track from a local file referenced in a multimedia file of the type described above. If the referenced file is not stored locally and the device is networked to the location where the device is stored, then the device can obtain a local copy of the file. The device would then access both files, establishing a video, an audio and a subtitle (if required) pipeline into which the various tracks of multimedia are fed from the different file sources.
Many embodiments of decoders in accordance with the present invention are capable of displaying a multimedia presentation contained within a multimedia file non-sequentially. Non-sequential display can include playing the sequence in reverse and/or increasing the apparent speed with which the sequence is displayed by skipping frames in the sequence. Non-sequential display can also include skipping in an irregular fashion between different portions of a multimedia presentation.
In several embodiments, the decoder uses an ‘index’ chunk within the ‘DXDT’ of a multimedia file to locate particular encoded video frames. Knowledge of the location of specific encoded video frames can be used to skip frames either in a regular fashion (such as during fast forwarding or rewinding) or in an irregular fashion (such as when skipping between scenes or chapters).
A process that can be used in accordance with an embodiment of the method of the invention to locate a specific video frame using an ‘index’ chunk is shown in
A process for locating the ‘tag’ chunk within an ‘index’ chunk that references the preceding frame closest to a desired video frame within a video sequence is shown in
When a next ‘tag’ chunk can be located from within the ‘index’ chunk, the process compares (143e) the positions of the encoded video frame referenced by the next ‘tag’ chunk and the desired video chunk. A decision (143f) is then made based upon whether the desired video frame lies between the encoded video frames in the sequence referenced by the current and next ‘tag’ chunks. If the desired video frame lies between the referenced frames, then the position of the video frame and any audio referenced by the current ‘tag’ chunk within the multimedia file are returned (143g) by the process.
When the desired video frame is not located between the frames referenced by the current and next ‘tag’ chunks, then the next ‘tag’ chunk becomes (143h) the current ‘tag’ chunk. The process repeats (143a) until the condition that the desired video frame be located between the encoded video frames referenced by the current and next ‘tag’ chunks is satisfied (143g) or all of the ‘tag’ chunks have been inspected (143d).
In further embodiments, the location of a desired frame can be further refined by using the references within the ‘index’ chunk to the corresponding information within the ‘idx1’ chunk to search for a reference within the ‘idx1’ chunk to a specifically desired encoded video frame that is not referenced in the ‘index’ chunk.
Although the processes shown in
A decoder in accordance with an embodiment of the present invention is illustrated in
An example of a menu in accordance with an embodiment of the present invention is illustrated in
As the user enters commands using the input device, the display may be updated not only in response to the selection of button areas but also simply due to the pointer being located within a button area. As discussed above, typically all of the media information used to generate the menus is located within the multimedia file and more specifically within a ‘DMNU’ chunk. Although in other embodiments, the information can be located elsewhere within the file and/or in other files.
‘Meta data’ is a standardized method of representing information. The standardized nature of ‘Meta data’ enables the data to be accessed and understood by automatic processes. In one embodiment, the ‘meta data’ is extracted and provided to a user for viewing. Several embodiments enable multimedia files on a server to be inspected to provide information concerning a users viewing habits and viewing preferences. Such information could be used by software applications to recommend other multimedia files that a user may enjoy viewing. In one embodiment, the recommendations can be based on the multimedia files contained on servers of other users. In other embodiments, a user can request a multimedia file and the file can be located by a search engine and/or intelligent agents that inspect the ‘meta data’ of multimedia files in a variety of locations. In addition, the user can chose between various multimedia files containing a particular multimedia presentation based on ‘meta data’ concerning the manner in which each of the different versions of the presentation were encoded.
In several embodiments, the ‘meta data’ of multimedia files in accordance with embodiments of the present invention can be accessed for purposes of cataloging or for creating a simple menu to access the content of the file.
While the above description contains many specific embodiments of the invention, these should not be construed as limitations on the scope of the invention, but rather as an example of one embodiment thereof. For example, a multimedia file in accordance with an embodiment of the present invention can include a single multimedia presentation or multiple multimedia presentations. In addition, such a file can include one or more menus and any variety of different types of ‘meta data’. Accordingly, the scope of the invention should be determined not by the embodiments illustrated, but by the appended claims and their equivalents.
The present application is a continuation of U.S. patent application Ser. No. 17/307,990 filed May 4, 2021, entitled “Multimedia Distribution System”, which application continuation of U.S. patent application Ser. No. 16/039,752 filed Jul. 19, 2018, entitled “Multimedia Distribution System” and issued on May 25, 2021 as U.S. Pat. No. 11,017,816, which application is a continuation of U.S. patent application Ser. No. 15/217,894 filed Jul. 22, 2016, entitled “Multimedia Distribution System” and issued on Jul. 24, 2018 as U.S. Pat. No. 10,032,485, which application is a continuation of U.S. patent application Ser. No. 13/913,187 filed Jun. 7, 2013, entitled “Multimedia Distribution System” and issued on Aug. 16, 2016 as U.S. Pat. No. 9,420,287, which application is a continuation of U.S. patent application Ser. No. 11/258,496 filed Oct. 24, 2005, entitled “Multimedia Distribution System” and issued on Jun. 25, 2013 as U.S. Pat. No. 8,472,792, which application is a continuation-in-part of U.S. patent application Ser. No. 11/016,184, filed on Dec. 17, 2004, entitled “Multimedia Distribution System” and issued on May 20, 2014 as U.S. Pat. No. 8,731,369, which application is a continuation-in-part of U.S. patent application Ser. No. 10/731,809, filed on Dec. 8, 2003, entitled “File Format for Multiple Track Digital Data” and issued on Apr. 14, 2009 as U.S. Pat. No. 7,519,274, and also claims priority from Patent Cooperation Treaty Patent Application No. PCT/US2004/041667, filed on Dec. 8, 2004, entitled “Multimedia Distribution System”, the disclosures of which are incorporated herein by reference in their entirety.
| Number | Date | Country | |
|---|---|---|---|
| Parent | 17307990 | May 2021 | US |
| Child | 17453314 | US | |
| Parent | 16039752 | Jul 2018 | US |
| Child | 17307990 | US | |
| Parent | 15217894 | Jul 2016 | US |
| Child | 16039752 | US | |
| Parent | 13913187 | Jun 2013 | US |
| Child | 15217894 | US | |
| Parent | 11258496 | Oct 2005 | US |
| Child | 13913187 | US |
| Number | Date | Country | |
|---|---|---|---|
| Parent | 11016184 | Dec 2004 | US |
| Child | 11258496 | US | |
| Parent | 10731809 | Dec 2003 | US |
| Child | 11016184 | US |
