Automatic camera angle switching in response to low noise audio to create combined audiovisual file

Abstract

A system and method are provided for automatically concatenating two or more audiovisual clips containing video input from multiple cameras, and producing a combined audiovisual file containing video that switches between the two video inputs. In some examples, two video inputs and an audio input are recorded synchronously and are synchronized. The audio input can be sampled to locate low-noise audio events. The audiovisual file contains video that switches between two or more camera angles at the low-noise audio events. In one aspect, pauses are automatically removed from the audiovisual files. In another aspect, the system detects switch-initiating events, and switches between camera angles in response to detecting a switch-initiating event.

Description

BACKGROUND

Video interviews can be taped and used by recruiters to assist in representing candidates to potential employers. These videos can sometimes be one-dimensional and uninteresting.

Videos that cut between multiple views of the candidate can be more visually interesting, but editing and producing high-quality video is tedious and time-consuming.

SUMMARY

A system and method for automatically producing audiovisual files containing video from multiple cameras is provided. In some examples, a system is provided having a first video input and a second video input; an audio input; a time counter providing a timeline associated with the first video input, the second video input, and the audio input, the timeline enables a time synchronization of the first video input, the second video input, and the audio input; a non-transitory computer memory and a computer processor; and computer instructions stored on the memory for instructing the processor to perform the steps of: sampling the audio input to identify a low noise audio segment in which the decibel level is below a threshold level for a predetermined period of time; and automatically assembling a combined audiovisual file by performing the steps of: retaining a first audiovisual clip can include a portion of the audio input and first video input occurring before the low noise audio segment, retaining a second audiovisual clip can include a portion of the audio input and second video input occurring after the low noise audio segment, and concatenating the first audiovisual clip and the second audiovisual clip to create a combined audiovisual file. In some examples, the first video input, the second video input, and the audio input are recorded synchronously, and the combined audiovisual file is a video interview of a job candidate.

In some examples, the first audiovisual clip ends at the low noise audio segment and the second audiovisual clip begins at the low noise audio segment. In some examples, the first audiovisual clip is earlier in the timeline than the second audiovisual clip, and the first audiovisual clip corresponds to a time immediately preceding the second audiovisual clip. In some examples, the predetermined period of time is at least two seconds. Some examples can further include computer instructions stored on the memory for instructing the processor to perform the steps of: sampling the audio input to identify a beginning of the low noise audio segment and an end of the low noise audio segment; removing portions of the audio input, the first video input, and the second video input that fall between the beginning and end of the low noise audio segment; and concatenating the first audiovisual clip and the second audiovisual clip to create a combined audiovisual file that does not contain the low noise audio segment; the first audiovisual clip includes a portion of the audio input and first video input occurring before the beginning of the low noise audio segment, and the second audiovisual clip includes a portion of the audio input and the second video input occurring after the end of the low noise audio segment.

In some examples, the low noise audio segment is at least four seconds long. Some examples further include computer instructions stored on the memory for instructing the processor to perform the steps of: sampling the audio input to identify multiple low noise audio segments in which the decibel level is below the threshold level for a predetermined period of time; and automatically concatenating alternating audiovisual clips that switch between the first video input and second video input after each low noise audio segment. Some examples further include computer instructions stored on the memory for instructing the processor to perform the steps of: sampling the audio input to identify multiple low noise audio segments in which the decibel level is below the threshold level for at least the predetermined period of time; extracting content data from the first video input, the second video input, or the audio input to identify one or more switch-initiating events; automatically assembling a combined audiovisual file that switches between the first video input and the second video input following a switch-initiating event. In some examples, the switch-initiating events include one or more of: a gesture recognition event; a facial recognition event; a length of time of at least 30 seconds since a most recent camera angle switch; or a keyword extracted from the audio input via speech-to-text.

In some examples, a computer-implemented method includes receiving first video input of an individual from a first camera, receiving second video input of the individual from a second camera, receiving audio input of the individual from a microphone, the first video input, the second video input, and the audio input are recorded synchronously; sampling the audio input, the first video input, or the second video input to identify an event; automatically assembling a combined audiovisual file by performing the steps of: retaining a first audiovisual clip can include a portion of the first video input occurring before the event; retaining a second audiovisual clip can include a portion of the second video input occurring after the event; and concatenating the first audiovisual clip and the second audiovisual clip to create a combined audiovisual file containing video of the individual from two camera angles.

In some examples, the combined audiovisual file is a video interview of a job candidate. In some examples, the event is a low noise audio segment. Some examples further include the steps of: sampling the audio input to identify a plurality of low noise audio segments; retaining video clips that alternately switch between the first video input and the second video input following the low noise audio segments; and concatenating the alternating video clips to create a combined audiovisual file containing video that alternates between two camera angles. Some examples further include the step of extracting content data from the first video input, the second video input, or the audio input to identify one or more switch-initiating events, switching between the first video input and the second video input is only performed for low noise audio segments that follow switch-initiating events.

In some examples, the content data is at least one of: facial recognition; gesture recognition; posture recognition; or keywords extracted using speech-to-text. Some examples further include the steps of: sampling the audio input to identify multiple extended low noise audio segments that are at least four seconds long; removing the portions of the audio input, the first video input, and the second video input that fall between the beginning and end of the extended low noise audio segments; concatenating video clips containing alternating portions of the first video input and portions of the second video input to create a combined audiovisual file that does not contain audio or video occurring between the beginning and end of extended low noise audio segments.

In some examples, a system is included having a first video input and a second video input; an audio input; a time counter providing a timeline associated with the first video input, the second video input, and the audio input, the timeline enables a time synchronization of the first video input, the second video input, and the audio input; a non-transitory computer memory and a computer processor; and computer instructions stored on the memory for instructing the processor to perform the steps of: sampling the audio input to identify a low noise audio segment in which the decibel level is below a threshold level for a predetermined period of time; and automatically assembling a combined audiovisual file by performing the steps of: retaining a first audiovisual clip can include a portion of the first video input and synchronized audio input occurring before the low noise audio segment; retaining a second audiovisual clip can include a portion of the second video input and synchronized audio input occurring after the low noise audio segment; and concatenating the first audiovisual clip and the second audiovisual clip to create a combined audiovisual file.

In some examples, the first video input, the second video input, and the audio input are recorded synchronously, and the combined audiovisual file is a video interview of a job candidate. Some examples further include computer instructions stored on the memory for instructing the processor to perform the steps of: sampling the audio input to identify a plurality of low noise audio segments in which the decibel level is below the threshold level for the predetermined period of time; and concatenating a plurality of audiovisual clips that switch between the first video input and the second video input after each low noise audio segment to create a combined audiovisual file containing video that alternates between two camera angles.

This summary is an overview of some of the teachings of the present application and is not intended to be an exclusive or exhaustive treatment of the present subject matter. Further details are found in the detailed description and appended claims. Other aspects will be apparent to persons skilled in the art upon reading and understanding the following detailed description and viewing the drawings that form a part thereof, each of which is not to be taken in a limiting sense.

BRIEF DESCRIPTION OF THE FIGURES

FIG. 1 is a perspective view of a multi-camera kiosk according to some examples.

FIG. 2 is a schematic view of a kiosk system according to some examples.

FIG. 3 illustrates an example of multiple video inputs.

FIG. 4 is a graph of decibel level versus time for an audio input according to some examples.

FIG. 5 visually illustrates a method of automatically concatenating audiovisual clips into an audiovisual file according to some examples.

FIG. 6 visually illustrates a method of removing pauses from audio and video inputs and automatically concatenating audiovisual clips into an audiovisual file according to some examples.

FIG. 7 visually illustrates a method of automatically concatenating audiovisual clips into an audiovisual file in response to an event according to some examples.

DETAILED DESCRIPTION

The present disclosure relates to a system and method for producing audiovisual files containing video that automatically cuts between video footage from multiple cameras. The multiple cameras can be arranged during recording such that they each focus on a subject from a different camera angle, providing multiple viewpoints of the subject. The system can be used for recording a person who is speaking, such as in a video interview. Although the system will be described in the context of a video interview, other uses are contemplated and are within the scope of the technology. For example, the system could be used to record educational videos, entertaining or informative speaking, or other situations in which an individual is being recorded with video and audio.

Some implementations of the technology provide a kiosk or booth that houses multiple cameras and a microphone. The cameras each produce a video input to the system, and the microphone produces an audio input. A time counter provides a timeline associated with the multiple video inputs and the audio input. The timeline enables video input from each camera to be time-synchronized with the audio input from the microphone.

Multiple audiovisual clips are created by combining video inputs with a corresponding synchronized audio input. The system detects events in the audio input, video inputs, or both the audio and video inputs, such as a pause in speaking corresponding to low-audio input. The events correspond to a particular time in the synchronization timeline. To automatically assemble audiovisual files, the system concatenates a first audiovisual clip and a second audiovisual clip. The first audiovisual clip contains video input before the event, and the second audiovisual clip contains video input after the event. The system can further create audiovisual files that concatenate three or more audiovisual clips that switch between particular video inputs after predetermined events.

One example of an event that can be used as a marker for deciding when to cut between different video clips is a drop in the audio volume detected by the microphone. During recording, the speaker may stop speaking briefly, such as when switching between topics, or when pausing to collect their thoughts. These pauses can correspond to a significant drop in audio volume. In some examples, the system looks for these low-noise events in the audio track. Then, when assembling an audiovisual file of the video interview, the system can change between different cameras at the pauses. This allows the system to automatically produce high quality, entertaining, and visually interesting videos with no need for a human editor to edit the video interview. Because the quality of the viewing experience is improved, the viewer is likely to have a better impression of a candidate or other speaker in the video. A higher quality video better showcases the strengths of the speaker, providing benefits to the speaker as well as the viewer.

In another aspect, the system can remove unwanted portions of the video automatically based on the contents of the audio or video inputs, or both. For example, the system may discard portions of the video interview in which the individual is not speaking for an extended period of time. One way this can be done is by keeping track of the length of time that the audio volume is below a certain volume. If the audio volume is low for an extended period of time, such as a predetermined number of seconds, the system can note the time that the low noise segment begins and ends. A first audiovisual clip that ends at the beginning of the low noise segment can be concatenated with a second audiovisual clip that begins at the end of the low noise segment. The audio input and video inputs that occur between the beginning and end of the low noise segment can be discarded. In some examples, the system can cut multiple pauses from the video interview, and switch between camera angles multiple times. This eliminates dead air and improves the quality of the video interview for a viewer.

In another aspect, the system can choose which video input to use in the combined audiovisual file based on the content of the video input. For example, the video inputs from the multiple cameras can be analyzed to look for content data to determine whether a particular event of interest takes place. As just one example, the system can use facial recognition to determine which camera the individual is facing at a particular time. The system then can selectively prefer the video input from the camera that the individual is facing at that time in the video. As another example, the system can use gesture recognition to determine that the individual is using their hands when talking. The system can selectively prefer the video input that best captures the hand gestures. For example, if the candidate consistently pivots to the left while gesturing, a right camera profile shot might be subjectively better than minimizing the candidate's energy using the left camera feed. Content data such as facial recognition and gesture recognition can also be used to find events that the system can use to decide when to switch between different camera angles.

In another aspect, the system can choose which video input to use based on a change between segments of the interview, such as between different interview questions.

Turning now to the figures, an example implementation of the disclosed technology will be described in relation to a kiosk for recording video interviews. However, it should be understood that this implementation is only one possible example, and other set ups could be used to implement the disclosed technology.

Video Interview Kiosk (FIG. 1)

FIG. 1 shows a kiosk 101 for recording a video interview of an individual 112. The kiosk 101 is generally shaped as an enclosed booth 105. The individual 112 can be positioned inside of the enclosed booth 105 while being recorded. Optionally, a seat 107 is provided for the individual 112. The kiosk 101 houses multiple cameras, including a first camera 122, a second camera 124, and a third camera 126. Each of the cameras is capable of recording video of the individual 112 from different angles. In the example of FIG. 1, the first camera 122 records the individual 112 from the left side, the second camera 124 records the individual 112 from the center, and the third camera 126 records the individual 112 from the right side. In some examples, the camera 124 can be integrated into a user interface 133 on a tablet computer 131. The user interface 133 can prompt the individual to answer interview questions. A microphone 142 is provided for recording audio.

The first, second, and third cameras 122, 124, 126 can be digital video cameras that record video in the visible spectrum using, for example, a CCD or CMOS image sensor. Optionally, the cameras can be provided with infrared sensors or other sensors to detect depth, movement, etc.

In some examples, the various pieces of hardware can be mounted to the walls of the enclosed booth 105 on a vertical support 151 and a horizontal support 152. The vertical support 151 can be used to adjust the vertical height of the cameras and user interface, and the horizontal support 152 can be used to adjust the angle of the cameras 122, 124, 126.

Schematic of Kiosk and Edge Server (FIG. 2)

FIG. 2 shows a schematic diagram of one example of the system. The kiosk 101 includes an edge server 201 that has a computer processor 203, a system bus 207, a system clock 209, and a non-transitory computer memory 205. The edge server 201 is configured to receive input from the video and audio devices of the kiosk and process the received inputs.

The kiosk 101 can further include the candidate user interface 133 in data communication with the edge server 201. An additional user interface 233 can be provided for a kiosk attendant. The attendant user interface 233 can be used, for example, to check in users, or to enter data about the users. The candidate user interface 133 and the attendant user interface 233 can be provided with a user interface application program interface (API) 235 stored in the memory 205 and executed by the processor 203. The user interface API 235 can access particular data stored in the memory 205, such as interview questions 237 that can be displayed to the individual 112 on in the user interface 133. The user interface API 235 can receive input from the individual 112 to prompt a display of a next question once the individual has finished answering a current question.

The system includes multiple types of data inputs. In one example, the camera 122 produces a video input 222, the camera 124 produces a video input 224, and the camera 126 produces a video input 226. The microphone 142 produces an audio input 242. The system also receives behavioral data input 228. The behavioral data input 228 can be from a variety of different sources. In some examples, the behavioral data input 228 is a portion of data received from one or more of the cameras 122, 124, 126. In other words, the system receives video data and uses it as the behavioral data input 228. In some examples, the behavioral data input 228 is a portion of data received from the microphone 142. In some examples, the behavioral data input 228 is sensor data from one or more infrared sensors provided on the cameras 122, 124, 126. The system can also receive text data input 221 that can include text related to the individual 112, and candidate materials 223 that can include materials related to the individual's job candidacy, such as a resume.

In some examples, the video inputs 222, 224, 226 are stored in the memory 205 of the edge server 201 as video files 261. In alternative examples, the video inputs 222, 224, 226 are processed by the processor 203, but are not stored separately. In some examples, the audio input 242 is stored as audio files 262. In alternative examples, the audio input 242 is not stored separately. The candidate materials input 223, text data input 221, and behavioral data input 228 can also be optionally stored or not stored as desired.

In some examples, the edge server 201 further includes a network communication device 271 that enables the edge server 201 to communicate with a remote network 281. This enables data that is received and/or processed at the edge server 201 to be transferred over the network 281 to a candidate database server 291.

The edge server 201 includes computer instructions stored on the memory 205 to perform particular methods. The computer instructions can be stored as software modules. As will be described below, the system can include an audiovisual file processing module 263 for processing received audio and video inputs and assembling the inputs into audiovisual files and storing the assembled audiovisual files 264. The system can include a data extraction module 266 that can receive one or more of the data inputs (video inputs, audio input, behavioral input, etc.) and extract behavior data 267 from the inputs and store the extracted behavior data 267 in the memory 205.

Automatically Creating Audiovisual Files from Two or More Video Inputs (FIGS. 3-7) The disclosed system and method provide a way to take video inputs from multiple cameras and arrange them automatically into a single audiovisual file that cuts between different camera angles to create a visually interesting product.

FIG. 3 illustrates video frames of video inputs received from different cameras. In this example, video frame 324 is part of the video input 224 that is received from the second camera 124, which focuses on the individual 112 from a front and center angle. This video input is designated as “Video 1” or simply “Vid1.” The video frame 322 is part of the video input 222 from the first camera 122, which focuses on the individual 112 from the individual 112's left side. This video input is designated as “Video 2” or simply “Vid2.” The video frame 326 is part of the video input 226 from the third camera 126, which focuses on the individual 112 from the individual 112's right side. This video input is designated as “Video 3” or simply “Vid3.” These video inputs can be provided using any of a number of different types of video coding formats. These include but are not limited to MPEG-2 Part 2, MPEG-4 Part 2, H.264 (MPEG-4 Part 10), HEVC, and AV1.

Audio inputs 242 can also be provided using any of a number of different types of audio compression formats. These can include but are not limited to MP1, MP2, MP3, AAC, ALAC, and Windows Media Audio.

The system takes audiovisual clips recorded during the video interview and concatenates the audiovisual clips to create a single combined audiovisual file containing video of an individual from multiple camera angles. In some implementations, a system clock 209 creates a timestamp associated with the video inputs 222, 224, 226 and the audio input 242 that allows the system to synchronize the audio and video based on the timestamp. A custom driver can be used to combine the audio input with the video input to create an audiovisual file.

As used herein, an “audiovisual file” is a computer-readable container file that includes both video and audio. An audiovisual file can be saved on a computer memory, transferred to a remote computer via a network, and played back at a later time. Some examples of video encoding formats for an audiovisual file compatible with this disclosure are MP4 (mp4, m4a, mov); 3GP (3gp, 3gp2, 3g2, 3gpp, 3gpp2); WMV (wmy, wma); AVI; and QuickTime.

As used herein, an “audiovisual clip” is a video input combined with an audio input that is synchronized with the video input. For example, the system can record an individual 112 speaking for a particular length of time, such as 30 seconds. In a system that has three cameras, three audiovisual clips could be created from that 30 second recording: a first audiovisual clip can contain the video input 224 from Vid1 synchronized with the audio input 242 from t=0 to t=30 seconds. A second audiovisual clip can contain the video input 222 from Vid2 synchronized with the audio input 242 from t=0 to t=30 seconds. A third audiovisual clip can contain the video input 226 from Vid3 synchronized with the audio input 242 from t=0 to t=30 seconds; Audiovisual clips can be created by processing a video input stream and an audio input stream which are then stored as an audiovisual file. An audiovisual clip as described herein can be, but is not necessarily stored in an intermediate state as a separate audiovisual file before being concatenated with other audiovisual clips. As will be described below, in some examples, the system will select one video input from a number of available video inputs, and use that video input to create an audiovisual clip that will later be saved in an audiovisual file. In some examples, the unused video inputs may be discarded.

Audiovisual clips can be concatenated. As used herein, “concatenated” means adding two audiovisual clips together sequentially in an audiovisual file. For example, two audiovisual clips that are each 30 seconds long can be combined to create a 60-second long audiovisual file. In this case, the audiovisual file would cut from the first audiovisual clip to the second audiovisual clip at the 30 second mark.

During use, each camera in the system records an unbroken sequence of video, and the microphone records an unbroken sequence of audio. An underlying time counter provides a timeline associated with the video and audio so that the video and audio can be synchronized.

In one example of the technology, the system samples the audio track to automatically find events that are used to triggered the system to cut between video inputs when producing an audiovisual file. In one example, the system looks for segments in the audio track in which the volume is below a threshold volume. These will be referred to as low noise audio segments.

FIG. 4 is a graph 411 representing the audio volume in an audio track over time. The graph conceptually shows the audio volume of the audio input in decibels (D) versus time in seconds (t). In some examples, the system uses a particular threshold volume as a trigger to determine when to cut between the video inputs. For example, in FIG. 4, the threshold level is 30 decibels. One method of finding low noise audio segments is to calculate an average decibel level over a particular range of time, such as 4 seconds. If the average decibel level during that period of time is below the threshold level, the system will mark this as a low noise audio segment.

Applying this method to FIG. 4, the system computes the average (mean) volume over each four-second interval for the entire length of the audio track, in this case, in the range between t=0 and t=35. Consider an average decibel level over a four second interval between t=5 and t=9. In this case, although the volume falls below 30 decibels for a short period of time, the average volume over that four second period is greater than 30 decibels, and therefore this would not be considered a low noise audio segment. Over the four second interval from t=11 to t=15 seconds, the average volume is less than 30 decibels, and therefore this would be considered a low noise audio segment. In some examples, as soon the system detects an event corresponding to a low noise audio segment, the system marks that time as being a trigger to switch between video inputs.

In some examples, the system marks the beginning and end of the low noise audio segments to find low noise audio segments of a particular length. In this example, the system computes the average (mean) volume over each four second interval, and as soon the average volume is below the threshold volume (in this case 30 decibels), the system marks that interval as corresponding to the beginning of the low noise audio segment. The system continues to sample the audio volume until the average audio volume is above the threshold volume. The system then marks that interval as corresponding to the end of the low noise audio segment.

The system uses the low noise audio segments to determine when to switch between camera angles. After finding and interval corresponding to the beginning or end of the low noise audio segments, the system determines precisely at which time to switch. This can be done in a number of ways, depending upon the desired result.

In the example of FIG. 4, the system could determine that the average volume of the four second interval between =10 and t=12 drops below the threshold volume. The system could use the end of that interval (t=12) to be the time to switch. Alternatively, the system could determine that the average volume of the four-second interval between t=18 and t=22 increases above the threshold volume, and determine that the beginning of that interval (t=18) as the time to switch. The system could also use the midpoint of the beginning and end of the intervals to switch (i.e., midway between t=12 and t=18). Other methods of determining precisely when in the timeline to make the switch are possible, and are within the scope of the technology.

In some examples, the system is configured to discard portions of the video and audio inputs that correspond to a portion of the low noise audio segments. This eliminates dead air and makes the audiovisual file more interesting for the viewer. In some examples, the system only discards audio segments that our at least a predetermined length of time, such as at least 2 seconds, at least 4 seconds, at least 6 seconds, at least 8 seconds, or at least 10 seconds. This implementation will be discussed further in relation to FIG. 6.

Automatically Concatenating Audiovisual Clips (FIG. 5)

FIG. 5 illustrates a system and method for automatically creating a combined audiovisual file containing video images from two or more video inputs. For the sake of simplicity, only two video inputs are illustrated in FIG. 5. It should be understood, however, that the method and system could be adapted to any number of video inputs.

The system includes two video inputs: Video 1 and Video 2. The system also includes an Audio input. In the example of FIG. 5, the video inputs and the audio input are recorded simultaneously. The two video inputs and the audio input are each recorded as an unbroken sequence. A time counter, such as the system clock 209, provides a timeline 501 that enables a time synchronization of the two video inputs and the audio input. The recording begins at time t₀ and ends at time t_n.

In the example of FIG. 5, the system samples the audio track to determine low noise audio segments. For example, the system can use the method as described in relation to FIG. 4; however, other methods of determining low noise audio segments are contemplated, and are within the scope of the disclosed technology.

Sampling the audio track, the system determines that at time t₁, a low noise audio event occurred. The time segment between t=t₀ and t=t₁ is denoted as Seg1. To assemble a combined audiovisual file 540, the system selects an audiovisual clip 541 combining one video input from Seg1 synchronized with the audio from Seg1, and saves this audiovisual clip 541 as a first segment of the audiovisual file 540—in this case, Vid1.Seg1 (Video 1 Segment 1) and Aud.Seg1 (audio Segment 1). In some examples, the system can use a default video input as the initial input, such as using the front-facing camera as the first video input for the first audiovisual clip. In alternative examples, the system may sample content received while the video and audio are being recorded to prefer one video input over another input. For example, the system may use facial or gesture recognition to determine that one camera angle is preferable over another camera angle for that time segment. Various alternatives for choosing which video input to use first are possible, and are within the scope of the technology.

The system continues sampling the audio track, and determines that at time t₂, a second low noise audio event occurred. The time segment between t=t₁ and t=t₂ is denoted as Seg2. For this second time segment, the system automatically switches to the video input from Video 2, and saves a second audiovisual clip 542 containing Vid2.Seg2 and Aud.Seg2. The system concatenates the second audiovisual clip 542 and the first audiovisual clip 541 in the audiovisual file 540.

The system continues sampling the audio track, and determines that at time t₃, a third low noise audio event occurred. The time segment between t=t₂ and t=t₃ is denoted as Seg3. For this third time segment, the system automatically cuts back to the video input from Video 1, and saves a third audiovisual clip 543 containing Vid1.Seg3 and Aud.Seg3. The system concatenates the second audiovisual clip 542 and the third audiovisual clip 543 in the audiovisual file 540.

The system continues sampling the audio track, and determines that at time t₄, a fourth low noise audio event occurred. The time segment between t=t₃ and t=t₄ is denoted as Seg4. For this fourth time segment, the system automatically cuts back to the video input from Video 2, and saves a fourth audiovisual clip 544 containing Vid2.Seg4 and Aud.Seg4. The system concatenates the third audiovisual clip 543 and the fourth audiovisual clip 544 in the audiovisual file 540.

The system continues sampling the audio track, and determines that no additional low noise audio events occur, and the video input and audio input stop recording at time t₁. The time segment between t=t₄ and t=t_n is denoted as Seg5. For this fifth time segment, the system automatically cuts back to the video input from Video 1, and saves a fifth audiovisual clip 545 containing Vid1.Seg5 and Aud.Seg5. The system concatenates the fourth audiovisual clip 544 and the fifth audiovisual clip 545 in the audiovisual file 540.

In some examples, audio sampling and assembling of the combined audiovisual file is performed in real-time as the video interview is being recorded. In alternative examples, the video input and audio input can be recorded, stored in a memory, and processed later to create a combined audiovisual file. In some examples, after the audiovisual file is created, the raw data from the video inputs and audio input is discarded.

Automatically Removing Pauses and Concatenating Audiovisual Clips (FIG. 6)

In another aspect of the technology, the system can be configured to create combined audiovisual files that remove portions of the interview in which the subject is not speaking. FIG. 6 illustrates a system and method for automatically creating a combined audiovisual file containing video images from two or more video input, where a portion of the video input and audio input corresponding to low noise audio segments are not included in the combined audiovisual file. For the sake of simplicity, only two video inputs are illustrated in FIG. 6. It should be understood, however, that the method and system could be adapted to any number of video inputs.

In the example of FIG. 6, the system includes a video input Video 1 and Video number two. The system also includes an Audio input. The video inputs and the audio input are recorded simultaneously in an unbroken sequence. A time counter, such as the system clock 209, provides a timeline 601 that enables a time synchronization of the two video inputs and the audio input. The recording begins at time to and ends at time t_n.

As in the example of FIG. 5, the system samples the audio track to determine low noise audio segments. In FIG. 6, the system looks for the beginning and end of low noise audio segments, as described above with relation to FIG. 4. Sampling the audio track, the system determines that at time t₁, a low noise audio segment begins, and at time t₂, the low noise audio segment ends. The time segment between t=t₀ and t=t₁ is denoted as Seg1. To assemble a combined audiovisual file 640, the system selects an audiovisual clip 641 combining one video input from Seg1 synchronized with the audio from Seg1, and saves this audiovisual clip 641 as a first segment of the audiovisual file 640 in this case, Vid1.Seg1 (Video 1 Segment 1) and Aud.Seg1 (audio Segment 1). The system then disregards the audio inputs and video inputs that occur during Seg2, the time segment between t=t₁ and t=t₂.

The system continues sampling the audio track, and determines that at time t₃, a second low noise audio segment begins, and at time t₄, the second low noise audio segment ends. The time segment between t=t₂ and t=t₃ is denoted as Seg3. For this time segment, the system automatically switches to the video input from Video 2, and saves a second audiovisual clip 642 containing Vid2.Seg3 and Aud.Seg3. The system concatenates the second audiovisual clip 642 and the first audiovisual clip 641 in the audiovisual file 640.

The system continues sampling the audio input to determine the beginning and end of further low noise audio segments. In the example of FIG. 6, Seg6 is a low noise audio segment beginning at time t₅ and ending at time t₆. Seg 8 is a low noise audio segment beginning at time t₇ and ending at time t₈. The system removes the portions of the audio input and video inputs that fall between the beginning and end of the low noise audio segments. At the same time, the system automatically concatenates retained audiovisual clips, switching between the video inputs after the end of each low noise audio segment. The system concatenates the audiovisual clips 643, 644, and 645 to complete the audiovisual file 640. The resulting audiovisual file 640 contains audio from segments 1, 3, 5, 7, and 9. The audiovisual file 640 does not contain audio from segments 2, 4, 6, or 8. The audiovisual file 640 contains alternating video clips from Video 1 and Video 2 that switch between the first video input and the second video input after each low noise audio segment.

Automatically Concatenating Audiovisual Clips with Camera Switching in Response to Switch-Initiating Events (FIG. 7)

In another aspect of the technology, the system can be configured to switch between the different video inputs in response to events other than low noise audio segments. These events will be generally categorized as switch-initiating events. A switch-initiating event can be detected in the content of any of the data inputs that are associated with the timeline. “Content data” refers to any of the data collected during the video interview that can be correlated or associated with a specific time in the timeline. These events are triggers that the system uses to decide when to switch between the different video inputs. For example, behavioral data input, which can be received from an infrared sensor or present in the video or audio, can be associated with the timeline in a similar manner that the audio and video images are associated with the timeline. Facial recognition data, gesture recognition data, and posture recognition data can be monitored to look for switch-initiating events. For example, if the candidate turns away from one of the video cameras to face a different video camera, the system can detect that motion and note it as a switch-initiating event. Hand gestures or changes in posture can also be used to trigger the system to cut from one camera angle to a different camera angle.

As another example, the audio input can be analyzed using speech to text software, and the resulting text can be used to find keywords that trigger a switch. In this example, the words used by the candidate during the interview would be associated with a particular time in the timeline.

Another type of switch-initiating event can be the passage of a particular length of time. A timer can be set for a number of seconds that is the maximum desirable amount of time for a single segment of video. For example, an audiovisual file can feel stagnant and uninteresting if the same camera has been focusing on the subject for more than 90 seconds. The system clock can set a 90 second timer every time that a camera switch occurs. If it is been greater than 90 seconds since the most recent switch-initiating event, expiration of the 90 second timer can be used as the switch-initiating event. Other amounts of time could be used, such as 30 seconds, 45 seconds, 60 seconds, etc., depending on the desired results.

Conversely, the system clock can set a timer corresponding to a minimum number of seconds that must elapse before a switch between two video inputs. For example, the system could detect multiple switch-initiating events in rapid succession, and it may be undesirable to switch back-and-forth between two video inputs too quickly. To prevent this, the system clock could set a timer for 30 seconds, and only register switch-initiating events that occur after expiration of the 30 second timer. Though resulting combined audiovisual file would contain audiovisual clip segments of 30 seconds or longer.

Another type of switch-initiating event is a change between interview questions that the candidate is answering, or between other segments of a video recording session. In the context of an interview, the user interface API 235 (FIG. 2) can display interview questions so that the individual 112 can read each interview question and then respond to it verbally. The user interface API can receive input, such as on a touch screen or input button, to indicate that one question has been answered, and prompt the system to display the next question. The prompt to advance to the next question can be a switch-initiating event.

Turning to FIG. 7, the system includes two video inputs: Video 1 and Video 2. The system also includes an Audio input. In the example of FIG. 7, the video inputs and the audio input are recorded simultaneously. The two video inputs and the audio input are each recorded as an unbroken sequence. A time counter, such as the system clock 209, provides a timeline 701 that enables a time synchronization of the two video inputs and the audio input. The recording begins at time to and ends at time t_n. In some examples, the system of FIG. 7 further includes behavioral data input associated with the timeline 701.

In the example of FIG. 7, the system automatically samples the audio input for low noise audio segments in addition to detecting switch-initiating events. The system can sample the audio input using the method as described in relation to FIG. 4; however, other methods of determining low noise audio segments are contemplated, and are within the scope of the disclosed technology.

In FIG. 7, the audio track is sampled in a manner similar to that of FIG. 5. The system determines that at time t₁, a low noise audio event occurred. The time segment between t=t₀ and t=t₁ is denoted as Aud.Seg1. However, no switch-initiating event was detected during Aud.Seg1. Therefore, unlike the system of FIG. 5, the system does not switch video inputs.

At time t₂, the system detects a switch-initiating event. However, the system does not switch between camera angles at time t₂, because switch-initiating events can occur at any time, including during the middle of a sentence. Instead, the system in FIG. 7 continues sampling the audio input to find the next low noise audio event. This means that a switch between two camera angles is only performed after two conditions have been met: the system detects a switch-initiating event, and then, after the switch-initiating event, the system detects a low noise audio event.

In some examples, instead of continuously sampling the audio track for low noise audio events, the system could wait to detect a switch-initiating event, then begin sampling the audio input immediately after the switch-initiating event. The system would then cut from one video input to the other video input at the next low noise audio segment.

At time t₃, the system determines that another low noise audio segment has occurred. Because this low noise audio segment occurred after a switch-initiating event, the system begins assembling a combined audiovisual file 740 by using an audiovisual clip 741 combining one video input (in this case, Video 1) with synchronized audio input for the time segment t=t₀ through t=t₃.

The system then waits to detect another switch-initiating event. In the example of FIG. 7, the system finds another low noise audio event at t₄, but no switch-initiating event has yet occurred. Therefore, the system does not switch to the second video input. At time t₅, the system detects a switch-initiating event. The system then looks for the next low noise audio event, which occurs at time t₆. Because time t₆ is a low noise audio event that follows a switch-initiating event, the system takes the audiovisual clip 742 combining video input from Video 2 and audio input from the time segment from t=t₃ to t=t₆. The audiovisual clip 741 is concatenated with the audiovisual clip 742 in the audiovisual file 740.

The system then continues to wait for a switch-initiating event. In this case, no switch-initiating event occurs before the end of the video interview at time t_n. The audiovisual file 740 is completed by concatenating an alternating audiovisual clip 743 containing video input from Video 1 to the end of the audiovisual file 740.

The various methods described above can be combined in a number of different ways to create entertaining and visually interesting audiovisual interview files. Multiple video cameras can be used to capture a candidate from multiple camera angles. Camera switching between different camera angles can be performed automatically with or without removing audio and video corresponding to long pauses when the candidate is not speaking. Audio, video, and behavioral inputs can be analyzed to look for content data to use as switch-initiating events, and/or to decide which video input to use during a particular segment of the audiovisual file. Some element of biofeedback can be incorporated to favor one video camera input over the others.

As used in this specification and the appended claims, the singular forms include the plural unless the context clearly dictates otherwise. The term “or” is generally employed in the sense of “and/or” unless the content clearly dictates otherwise. The phrase “configured” describes a system, apparatus, or other structure that is constructed or configured to perform a particular task or adopt a particular configuration. The term “configured” can be used interchangeably with other similar terms such as arranged, constructed, manufactured, and the like.

All publications and patent applications referenced in this specification are herein incorporated by reference for all purposes.

While examples of the technology described herein are susceptible to various modifications and alternative forms, specifics thereof have been shown by way of example and drawings. It should be understood, however, that the scope herein is not limited to the particular examples described. On the contrary, the intention is to cover modifications, equivalents, and alternatives falling within the spirit and scope herein.

Claims

1. A system comprising: a first video input from a first video camera and a second video input from a second video camera, wherein the first video camera and the second video camera are directed towards a common area, such that the video cameras are configured to obtain video data of a common subject in the common area from different angles, wherein the first video input and the second video input are recorded simultaneously and are synchronized;an audio input, wherein the audio input is recorded simultaneously and is synchronized with the first video input and the second video input;a non-transitory computer memory and a computer processor; and computer instructions stored on the memory for instructing the processor to perform the steps of: sampling the audio input to identify a first low noise audio segment in which a decibel level is below a threshold level for a predetermined period of time and to identify a beginning of the first low noise audio segment;sampling the audio input to identify a second low noise audio segment in which a decibel level is below the threshold level for the predetermined period of time and to identify a beginning of the second low noise audio segment;determining an amount of time between the first low noise audio segment and a second low noise audio segment;producing a combined audiovisual production, wherein the combined audiovisual production comprises: a first audiovisual clip, the first audiovisual clip comprising: a portion of the audio input, anda portion of the first video input occurring before the first low noise audio segment, wherein the first audiovisual clip ends after the beginning of the first low noise audio segment and before an end of the first low noise audio segment,a second audiovisual clip, the second audio visual clip comprising: a portion of the audio input, anda portion of the second video input occurring immediately after an end of the portion of the first video input in the first audiovisual clip, the second audiovisual clip begins after the beginning of the first low noise audio segment and before the end of the first low noise audio segment, anda third audiovisual clip, the third audiovisual clip comprising: a portion of the audio input, and(1) if the amount of time between the first low noise audio segment and the second low noise audio segment is greater than a switch delay time period, a portion of the first video input occurring immediately after an end of the second video input in the second audiovisual clip, the third audiovisual clip begins after the beginning of the second low noise audio segment and before the end of the second low noise audio segment, or(2) if the amount of time between the first low noise audio segment and the second low noise audio segment is not greater than the switch delay time period, a portion of the second video input occurring immediately after an end of the second video input in the second audiovisual clip, the third audiovisual clip begins after the beginning of the second low noise audio segment and before the end of the second low noise audio segment.

2. The system of claim 1, wherein the first video input, the second video input, and the audio input are recorded synchronously.

3. The system of claim 1, wherein the combined audiovisual production is a video interview of a job candidate.

4. The system of claim 1, further comprising computer instructions stored on the memory for instructing the processor to perform the steps of: sampling the audio input to identify multiple low noise audio segments in which the decibel level is below the threshold level for a predetermined period of time; andautomatically concatenating alternating audiovisual clips that switch between the first video input and second video input after each low noise audio segment.

5. The system of claim 1, further comprising computer instructions stored on the memory for instructing the processor to perform the steps of: extracting content data from the first video input, the second video input, or the audio input to identify one or more switch-initiating events;automatically assembling a combined audiovisual production that switches between the first video input and the second video input following a switch-initiating event.

6. The system of claim 5, wherein the switch-initiating event is a keyword extracted from the audio input via speech-to-text.

7. The system of claim 1, wherein the combined audiovisual production has a length of time equivalent to a length of time of the first video input, the second video input and the audio input.

8. The system of claim 1, wherein when the first audiovisual clip ends after the beginning of the first low noise audio segment and before the end of the first low noise audio segment and the second audiovisual clip begins after the beginning of the first low noise audio segment and before the end of the first low noise audio segment.

9. A computer-implemented method comprising: receiving first video input from a first video camera;receiving second video input from a second video camera, wherein the first video camera and the second video camera are directed to a common area, such that the video cameras are configured to obtain video data of a common subject in the common area from different angles;receiving audio input from a microphone, wherein the first video input, the second video input and the audio input are recorded simultaneously and are synchronized;sampling the audio input to identify a first low noise audio segment, wherein the first low noise audio segment has a decibel level that is below a threshold level for a period of time, and to identify a second low noise audio segment, wherein the second low noise audio segment has a decibel level that is below the threshold level for the period of time;outputting an audiovisual production, wherein the audiovisual production comprises: a first audiovisual clip, the first audiovisual clip comprising: a portion of the audio input, anda portion of the first video input occurring before the first low noise audio segment, wherein the first audiovisual clip ends after the beginning of the first low noise audio segment and before an end of the first low noise audio segment,a second audiovisual clip, the second audio visual clip comprising: a portion of the audio input, anda portion of the second video input occurring immediately after an end of the portion of the first video input in the first audiovisual clip, the second audiovisual clip begins after the beginning of the first low noise audio segment and before the end of the first low noise audio segment, anda third audiovisual clip, the third audiovisual clip comprising: a portion of the audio input, and(1) if the amount of time between the first low noise audio segment and the second low noise audio segment is greater than a switch delay time period, a portion of the first video input occurring immediately after an end of the second video input in the second audiovisual clip, the third audiovisual clip begins after the beginning of the second low noise audio segment and before the end of the second low noise audio segment, or(2) if the amount of time between the first low noise audio segment and the second low noise audio segment is not greater than the switch delay time period, a portion of the second video input occurring immediately after an end of the second video input in the second audiovisual clip, the third audiovisual clip begins after the beginning of the second low noise audio segment and before the end of the second low noise audio segment.

10. The method of claim 9, wherein the output audiovisual production is a video interview of a job candidate.

11. The method of claim 9, wherein the first video input, the second video input, and the audio input are recorded synchronously.

12. The method of claim 9, further comprising the steps of: retaining video clips that alternately switch between the first video input and the second video input following the low noise audio segments the occur at least more than the switch delay time period after the previous low noise audio segment; andoutputting the alternating video clips to create an audiovisual production containing video that alternates between two camera angles.

13. The method of claim 9, further comprising the step of extracting content data from the first video input, the second video input, or the audio input to identify one or more switch-initiating events, wherein switching between the first video input and the second video input is only performed for low noise audio segments that follow switch-initiating events.

14. The method of claim 13, wherein the content data is a keyword extracted using speech-to-text.

15. The method of claim 9, wherein the audiovisual production comprises the first video input before the first low noise audio segment and during at least a portion of the first low noise audio segment, and the second video input during at least a portion of the first low noise audio segment and after the first low noise audio segment.

16. The method of claim 9, further comprising: discarding at least a portion of the first video input, the second video input, and the audio input of the low noise audio event when the low noise audio event exceeds a predetermined length of time, such that the output audiovisual production is of shorter than the received video inputs and audio input.

17. A system comprising: a first video input from a first video camera, a second video input from a second video camera, a third video input from a third video camera, wherein the first video camera, the second video camera, and the third video camera are directed towards a common area, such that the video cameras are configured to obtain video data of a common subject in the common area from different angles, wherein the first video input, second video input, and the third video input are recorded simultaneously and are synchronized;an audio input, wherein the audio input is recorded simultaneously and is synchronized with the first video input, the second video input, and the third video input;a non-transitory computer memory and a computer processor; and computer instructions stored on the memory for instructing the processor to perform the steps of: monitoring the audio input and the video inputs for switch-initiating events; andoutputting an audiovisual production;wherein the audiovisual production comprises: a first audiovisual clip, the first audiovisual clip comprising: a portion of the audio input occurring before a first switch-initiating event, anda portion of the first video input occurring before a first switch-initiating event, wherein the first audiovisual clip ends after a beginning of the first switch-initiating event;a second audiovisual clip, the second audiovisual clip comprising: a portion of the audio input occurring after the first switch-initiating event and before a second switch-initiating event, anda portion of the second video input occurring after the first switch-initiating event and before the second switch-initiating event, wherein the second audiovisual clip ends after a beginning of the second switch-initiating event; anda third audiovisual clip, the third audiovisual clip comprising: a portion of the audio input occurring after the second switch-initiating event, and(1) if the amount of time between the first switch-initiating event and the second switch-initiating event is greater than a switch delay time period, a portion of the third video input occurring immediately after an end of the second video input in the second audiovisual clip, the third audiovisual clip begins after the beginning of the second switch-initiating event, or(2) if the amount of time between the first switch-initiating event and the second switch-initiating event is not greater than the switch delay time period, a portion of the second video input occurring immediately after the end of the second video input in the second audiovisual clip, the third audiovisual clip begins after the beginning of the second switch-initiating event.

18. The system of claim 17, wherein the switch-initiating event comprises a keyword extracted from the audio input via speech-to-text.

19. The system of claim 17, wherein the switch-initiating event comprises a length of time of at least 30 seconds since a most recent camera angle switch or a keyword extracted from the audio input via speech-to-text.

20. The system of claim 17, wherein each video camera is directed towards a common area, such that each of the video cameras are configured to obtain video data of a common subject in the common area from different angles.