Patent Application
20170330540
A portion of the disclosure of this patent document contains material, which is subject to copyright protection. The copyright owner has no objection to the facsimile reproduction by anyone of the patent document or the patent disclosure, as it appears in the Patent and Trademark Office patent file or records, but otherwise reserves all copyright rights whatsoever.
This application claims priority to U.S. Patent Application No. 62/334,458 filed May 11, 2016; the disclosure of which is incorporated by reference in its entirety. This application is also related to U.S. Pat. No. 9,076,346 issued Jul. 7, 2015; the disclosure of which is incorporated by reference in its entirety.
The present invention relates generally to music theory, musical instruments, and the making, performance, selection, and playback of music. Particularly, the present invention relates to the selection of songs and making of song playlists or mixtapes based on the acoustic features of the songs.
In recent years with most recorded music available in digital form, it became vastly accessible and convenient to average consumers to own or have access to large number of songs anytime anywhere. This in turn necessitates sorting and categorizing song files into collections, playlists, or mixtapes. Besides manual operations by users, many electronic devices and software programs, such as Apple's iTunes, Pandora, and Spotify, now existed to automatically sort and categorize a plurality of songs based on song attributes, user preferences, recommendations from group of multiple users, etc. Some of these electronic devices and software programs can also make automatic recommendations for inclusion of new songs in collections, playlists, or mixtapes by predicting the user's music tastes.
The most commonly used sorting and categorizing criteria include song attributes such as song artist, composer, release date, genre, country of origin; attributes on the user's listening habits such as play frequency, play count, and last play time and date; and attributes on listening habits of a community of users such as aggregated group rating on the song, popularity, and demographic of the user group. However, all of these criteria are based on extrinsic or predefined data of the song and nothing on the substance of the song itself. It is not known that any existing technology allows song sorting and categorizing by the acoustic features of the songs, which could enable sorting and categorizing criteria based on, for example, the mood (e.g. happy, sad, etc.) of the songs.
The present invention provides a method and an apparatus to generate a song collection, playlist, or mixtape using sorting and categorizing criteria that are based on one or more acoustic features of the songs. In one embodiment, the method comprises 1.) importing the audio digital data files (e.g. MP3 files) from a library or repository of songs; 2.) sorting and filtering the audio digital data files based on a set of selection criteria, which can be translated into one or more acoustic feature parameter values; and 3.) creating a song collection, playlist, or mixtape based on the sorted and filtered results, attaching a graphical cover art and title for the song collection, playlist, or mixtape, wherein the graphical cover art and the title may be suggested based on an analysis of the characteristics of the songs in the song collection, playlist, or mixtape.
Embodiments of the invention are described in more detail hereinafter with reference to the drawings, in which:
In the following description, methods and systems for making song selection and creating song collections, playlists, or mixtapes are set forth as preferred examples. It will be apparent to those skilled in the art that modifications, including additions and/or substitutions may be made without departing from the scope and spirit of the invention. Specific details may be omitted so as not to obscure the invention; however, the disclosure is written to enable one skilled in the art to practice the teachings herein without undue experimentation.
Referring to
Referring to
The spectral analysis of an audio digital data file analyzes the data in each frame of the audio digital data file and extract the low level acoustic feature parameter values. The values of each low level acoustic feature parameter extracted over the entire set of frames (e.g. whole song) are then used to compute at least the minimum, maximum, variance, skewness, mean of derivative, variance of derivative, mean of second derivative, and variance of second derivative. Table 1 below listed the low level acoustic feature parameters extracted:
From the above low level acoustic feature parameter minimum, maximum, variance, skewness, mean of derivative, variance of derivative, mean of second derivative, and variance of second derivative values, high level feature parameters are modeled. In one embodiment, the high level acoustic feature parameters include those listed in Table 2 below.
In one embodiment of the high level acoustic feature parameter modeling, the value of each of the high level feature parameter is computed using statistical aggregation, which can be based on a weighted sum of one or more low level acoustic feature parameter's minimum, maximum, variance, skewness, mean of derivative, variance of derivative, mean of second derivative, and variance of second derivative values. For example, Danceability is determined by:
D=C
AL*Average Loudness's maximum+CBC*Beats Count's minimum+CZCR*Zero Cross Rate's mean; where:
CAL is the weight coefficient for Average Loudness;
CBC is the weight coefficient for Beats Count; and
CZCR is the weight coefficient for Zero Cross Rate; and
if D>a threshold value, then Danceability is true.
The extracted low level acoustic feature parameter values and the high level acoustic feature parameter values of the analyzed song are indexed and stored in a song acoustic feature database with reference to the analyzed song. With the songs in the accessible library or repository continuously being imported and analyzed, their acoustic feature parameter values extracted, computed, indexed and stored. This allows the rapid retrieval of the acoustic feature parameter values of any song in the library or repository during the composition of a song collection, playlist, or mixtape. This in turn speeds up the song sorting and categorizing response time perceived by the user.
The subsequent sorting and filtering of the audio digital data files by matching the acoustic feature parameters' similarity score values to the selection criteria comprise receiving a user input indicating the sorting and filtering criteria for compiling the song collection, playlist, or mixtape. In one embodiment, the user input includes, but not limited to, scaled values of tempo, intensity, bass level, treble level, rhythm, mood, and energy. These scaled values are then interpreted into desired acoustic feature parameter values. In one embodiment, only high level acoustic feature parameter values are interpreted; in other embodiments, it is possible that both low level and high level acoustic feature parameter values are interpreted. The analyzed songs' acoustic feature parameter values stored in the song acoustic feature database are retrieved to compare against the desired acoustic feature parameter values. The comparison is conducted by computing a similarity score for each song by a trained mean squared errors (MSE) function defined by:
where:
s is the similarity score;
n is the total number of acoustic feature compared;
ci is the importance coefficient for the acoustic feature i;
xi is the song's acoustic feature i parameter value; and
yi is the desired acoustic feature i parameter value.
Thus, the sorting of the analyzed songs is such that the higher the similarity score, the higher the ranking of the song in the compiled song collection, playlist, or mixtape. Further, a filter threshold can be used to filter out songs with similarity scores lower than the filter threshold.
To further enhance the speed performance, a similarity matrix can be pre-compiled and stored in the song acoustic feature database. In the compilation of the similarity matrix, each analyzed song having a corresponding data record of acoustic feature parameter values is compared against another to obtain a similarity score. In one embodiment, a data record of the similarity matrix may take the form of that shown in Table 3 below.
where xi identifies one analyzed song and xj identifies a different analyzed song. With the similarity matrix, once an analyzed song is added to a song collection, playlist, or mixtape, other candidate songs having similar acoustic features can be identified quickly by looking up the similarity matrix. To optimize the size of the similarity matrix, song pairs that are too dissimilar (e.g. below a similarity threshold) are excluded. Also, song pairs with identical acoustic features are also excluded.
With the list of sorted and filtered songs, a song collection, playlist, or mixtape is formed. In one embodiment, the graphical cover art to be attached to the song collection, playlist, or mixtape can be selected manually by user. In another embodiment, it can be generated automatically by a cover art generator. The cover art generator takes one or more image generation seeds (e.g. random pixel values) and one or more statistical aggregations of the constituent songs' acoustic feature parameter (e.g. mood) values as input and uses them as coefficients to generate a fractal-based image. In yet another embodiment, the graphical cover art to be attached can be recommended based on the results from an image search query in an image database having a plurality of graphics art images stored there within or the Internet using one or more constituent songs' metadata such as song artist name, song title, and release information. The concatenation of whole or parts of the metadata of more than one constituent songs may be used to form the image search key word.
A title picker may be used to recommend one or more titles for the compiled song collection, playlist, or mixtape using one or more concatenations of two or more constituent songs' artist names and titles. The recommended titles may also take reference to the prominent acoustic feature parameters of the constituent songs. In this case, the title picker detects the tag-style output of the high level acoustic feature parameter values to construct mood-based titles (e.g. “Happy” and “Sad”), and then combine them with other metadata of the constituent songs or contextual information, such as the time of day or weather to generate the title recommendations.
In accordance to one embodiment of the present invention, the method of making song selection and creating song collections, playlists, or mixtapes is implemented by a system shown in
The electronic embodiments disclosed herein may be implemented using general purpose or specialized computing devices, computer processors, or electronic circuitries including but not limited to application specific integrated circuits (ASIC), field programmable gate arrays (FPGA), and other programmable logic devices configured or programmed according to the teachings of the present disclosure. Computer instructions or software codes running in the general purpose or specialized computing devices, computer processors, or programmable logic devices can readily be prepared by practitioners skilled in the software or electronic art based on the teachings of the present disclosure.
All or portions of the electronic embodiments may be executed in one or more general purpose or computing devices including server computers, personal computers, laptop computers, mobile computing devices such as “smartphones” and “tablet” computers.
The electronic embodiments include computer storage media having computer instructions or software codes stored therein which can be used to program computers or microprocessors to perform any of the processes of the present invention. The storage media can include, but are not limited to, floppy disks, optical discs, Blu-ray Disc, DVD, CD-ROMs, and magneto-optical disks, ROMs, RAMs, flash memory devices, or any type of media or devices suitable for storing instructions, codes, and/or data.
The foregoing description of the present invention has been provided for the purposes of illustration and description. It is not intended to be exhaustive or to limit the invention to the precise forms disclosed. Many modifications and variations will be apparent to the practitioner skilled in the art.
The embodiments were chosen and described in order to best explain the principles of the invention and its practical application, thereby enabling others skilled in the art to understand the invention for various embodiments and with various modifications that are suited to the particular use contemplated.
| Number | Date | Country | |
|---|---|---|---|
| 62334458 | May 2016 | US |
