GB2508115A

GB2508115A - Generating audio signatures and providing an indication of lip sync delay

Info

Publication number: GB2508115A
Application number: GB1403468.0A
Authority: GB
Inventors: Pascal Carrieres
Original assignee: Miranda Technologies Inc
Current assignee: Miranda Technologies Inc
Priority date: 2009-11-30
Filing date: 2010-11-23
Publication date: 2014-05-21
Anticipated expiration: 2030-11-23
Also published as: GB2508115A8; GB2511655A; GB2511655B; HK1198311A1; GB201403468D0; WO2011063520A1; GB201407414D0; GB201210285D0; GB2489133A; GB2508115B; HK1176203A1; GB2489133B

Abstract

A method for generating an audio signature, representing audio signal content, comprises: performing first filtering 204 of a received audio signal, providing a first filtered audio signal; performing second audio signal filtering 206, providing a second filtered audio signal; comparing 208 the first to the second filtered audio signal; and assigning 210 a value depending on the comparison, generating the audio signature representing the content. Also claimed is an apparatus (Figure 5) providing a lip sync indication, comprising: a first signature extraction unit, receiving first audio and first video signals, providing a first video and first audio signature signals; a second extraction unit receiving second audio and second video signals, providing second video and audio signature signals; and a signature analysis unit receiving the various first and second video and audio signature signals. The first and second video signature signals are correlated to provide a video delay; further, the first with the second audio signature signal are correlated to provide an audio delay. The signature analysis unit further compares the audio and video delays and provides a lip sync indication if audio delay is different from video delay.

Description

METHOD AND APPARATUS FOR PROVIDING SIGNATURES OF

AUDIONIDEO SIGNALS AND FOR MAKING USE THEREOF

CROSS-REFERENCE AND RELATED APPLICATIONS

This application claims priority of US patent application No 12/627,728 that was filed on November 30, 2009 and entitled "METHOD AND APPARATUS FOR PROVIDING SIGNATURES OF AUDIO/VIDEO SIGNALS AND FOR MAKING USE THEREOF", the specification of which is hereby incorporated by reference.

This application further claims priority of Canadian patent application No 2,686,869 that was filed on November 30, 2009 and entitled "METHOD AND APPARATUS FOR PROVIDING SIGNATURES OF AUDIO/VIDEO SIGNALS AND FOR MAKING USE THEREOF", the specification of which is hereby incorporated by reference.

FIELD OF THE INVENTION

The invention relates to the processing of audio and video signals. More precisely, this invention pertains to a method and apparatus for providing signatures of audio/video signals and to a method and apparatus for using the provided signatures for detecting lip sync.

BACKGROUND OF THE INVENTION

Being able to detect differences in signals is of great interest for the skilled addressee. In the case of audio/video signals, differences may occur due to various reasons.

For instance, an audio signal may become desynchronized with a corresponding video signal during the course of a transmission of the audio/video signal. Such phenomenon is known by the skilled addressee as lip sync. Detecting lip sync may be of great interest for multiple system operators (MSO) that are looking for example to improve the quality of their channels and to avoid a phenomenon unacceptable at a certain degree for the end user.

Unfortunately, many prior art references for detecting lip sync disclose techniques that are either cumbersome to implement or require excessive processing resources.

Of interest though is PCT application WO 2008/066930 which discloses extracting features of audio and video signal content to provide reliable identification of the signals. In this international application a plurality of intermediate values are extracted. Those intermediate values represent one or more measures of dissimilarities between first and second values. Signatures are then generated by applying a hashing function on those intermediates values.

Unfortunately, the processing associated with such signature generation requires extensive processing capabilities which may not be acceptable in the case of certain applications wherein fast processing and limited processing capabilities are required.

There is therefore a need for a method and apparatus that will overcome at least one of the above-identified drawbacks.

Features of the invention will be apparent from review of the disclosure, drawings

and description of the invention below.

BRIEF SUMMARY

It is an object of the invention to provide a method and apparatus for generating a video signature representative of a content of a video signal.

The content of a video signal should be understood as an indication of an evolution of the content of the video signal over time.

Yet another object of the invention is to provide a method and apparatus for generating a signature representative of a content of an audio signal.

Similarly, the content of an audio signal should be understood as an indication of an evolution of the content of the audio signal over time.

The invention provides a method for generating a video signature representative of a content of a video signal, the method comprising receiving pixel data of each of a plurality of given pixels in a first image of the video signal, receiving pixel data of each of the plurality of given pixels in a second image subsequent to the first image, for each given pixel of the plurality of given pixels, comparing corresponding pixel data of the given pixel of the first image with the corresponding pixel data of the given pixel in the second image to provide a corresponding indication of a difference between the pixel data of the given pixel of the first image and the pixel data of the given pixel in the second image; if the difference is greater than a threshold, incrementing a counter and providing an indication of the counter to thereby generate the video signature.

An advantage of the method disclosed is that such method may be used for providing information, i.e. a signature, for characterizing the content of the video signal. When signatures are used for characterizing each of two video signals, those signatures may be advantageously correlated and in turn be used for detecting a delay between the two video signals which is of great advantage.

Another advantage of the method disclosed is that it requires limited processing for generating the signature which is therefore of great advantage.

In accordance with an embodiment, the method further comprises selecting a plurality of given pixels according to at least one criteria.

In accordance with an embodiment, the at least one criteria comprises a spatial criteria.

In yet another embodiment the threshold is equal to thirty two (32).

In another embodiment of the method, the providing of the indication of the counter comprises normalizing the indication of the counter.

The invention further provides a method for generating an audio signature representative of a content of an audio signal, the method comprising receiving an audio signal; performing a first filtering of the audio signal to provide a first filtered audio signal; performing a second filtering of the audio signal to provide a second filtered audio signal; comparing the first filtered audio signal to the second filtered audio signal and assigning a value depending on the comparison to thereby generate the audio signature representative of the content of the audio signal.

An advantage of the method disclosed is that such method may be used for providing information, i.e. a signature, for characterizing the content of the audio signal. When signatures are used for characterizing each of two audio signals, those two signatures may be advantageously correlated and in turn be used for detecting a delay between the two audio signals which is of great advantage.

Other advantages of the method disclosed above are that the method requires limited processing resources and limited storing capabilities for generating the signature which is of great advantage over the prior ad.

An advantage of the method disclosed is that such method may be used for detecting lip sync between a first audio/video signal and a second audio/video signal. By computing a first delay between the first audio signal and the second audio signal and a second delay between the first video signal and the second video signal and comparing the first and second delays it is possible to obtain an indication of the lip sync, if applicable.

In one embodiment of the method, the first filtering comprises detecting an envelope signal of the received audio signal.

In yet another embodiment of the method, the second filtering comprises computing an average value of the received audio signal.

In yet another embodiment, the assigning of a value depending on the comparison to thereby generate the audio signature representative of the content of the audio signal comprises performing decimation.

There is further disclosed an audio signature extraction unit for providing an audio signature signal AS representative of a content of an audio signal A,, the audio signature extraction unit comprising a first filtering unit for receiving the audio signal A and for providing a first filtered signal; a second filtering unit for receiving the audio signal A, and providing a second filtered signal; a comparator for receiving the first filtered signal and the second filtered signal and for performing a comparison of the first filtered with the second filtered signal and for providing a comparison result to thereby provide the audio signature signal AS representative of a content of an audio signal.

In one embodiment of the audio signature extraction unit, the first filtering unit comprises an envelope detector for providing a signal E indicative of an envelope of the audio signal A. In another embodiment of the audio signature extraction unit, the second filtering unit comprises a mean detector for providing a signal M indicative of a mean the audio signal A,.

In yet another embodiment of the audio signature extraction unit, the audio signature extraction unit further comprises an absolute value providing unit for receiving the audio signal 4 and for providing a signal indicative of an absolute value of the value of the audio signal A; wherein each of the mean detector and the envelope detector receives the signal indicative of an absolute value of the value of the audio signal 4.

In yet another embodiment of the audio signature extraction unit, the audio signature extraction unit further comprises a decimator for receiving the audio signature signal AS, and for decimating the audio signature signal AS,.

There is further disclosed a video signature extraction unit for providing a video signature signal representative of a content of an incoming image signal the video signature extraction unit comprising a first image pixel data providing unit for receiving the incoming image signal and for providing a corresponding pixel data of a pixel k of a plurality of given pixels 1 to N of a first image; a second image pixel data providing unit for receiving the incoming image signal and for providing a corresponding pixel data of the corresponding given pixel k in a second image; a comparator for receiving the corresponding pixel data and the corresponding pixel data Lk and for comparing the corresponding pixel

P C

data Ic with the corresponding pixel data k to provide an a result signal indicative of the comparison; a counter for receiving and counting the result signal of each of the plurality of given pixels 1 to N and for providing a counter signal and a video signature providing unit for receiving the counter signal and for providing the video signature signal In one embodiment of the video signature extraction unit, the result signal is a logic value one (1) if Abs(T -Cfr) is greater than thirty two (32) and a logic value zero (0) otherwise.

In yet another embodiment of the video signature extraction unit, the video signature providing unit receives the counter signal and divides the counter signal by N/240, wherein N is the number of given pixels.

In yet another embodiment of the video signature extraction unit, the video extraction unit further comprises a filtering unit for receiving the incoming image signal I and for providing a corresponding filtered signal to the first image pixel data providing unit and to the second image pixel data providing unit.

In yet another embodiment of the video signature extraction unit, the video extraction unit further comprises a windowing unit, wherein the incoming image signal fr is received by the windowing unit an a corresponding selected signal selected signal W is provided to the filtering unit.

In yet another embodiment of the video signature extraction unit, the video signature extraction unit further comprises a windowing unit for receiving the incoming image signal k and for providing a selected signal H'.

There is further provided an apparatus for providing an indication of a lip sync, the apparatus comprising a first audio signature extraction unit for receiving a first audio signal and a first video signal and for providing a first video signature and a first audio signature; a second signature extraction unit for receiving a second audio signal and a second video signal and for providing a second video signature and a second audio signature and a signature analysis unit for receiving the first video signature, the first audio signature, the second video signature and the second audio signature and for correlating the first video signature with the second video signature to provide a video delay and for further correlating the first audio signature with the second audio signature to provide an audio delay, the signature analysis unit for further comparing the audio delay with said video delay and for providing an indication of a lip sync is the audio delay is different from said video delay.

In one embodiment of the apparatus for providing an indication of a lip sync, the first signature extraction unit comprises a first audio signature extraction unit for receiving the first audio signal and for providing the first audio signature signal and the first signature extraction unit comprises a first video signature extraction unit for receiving the first video signal and for providing the first video signature signal.

In yet another embodiment of the apparatus for providing an indication of a lip sync, the second signature extraction unit comprises a second audio signature extraction unit for receiving the second audio signal and for providing the second audio signature signal and the second signature extraction unit comprises a second video signature unit for receiving the second video signal and for providing the second video signature signal.

In yet another embodiment of the apparatus for providing an indication of a lip sync, the first audio signature extraction unit comprises a first filtering unit for receiving the first audio signal and for providing a first filtered signal, a second filtering unit for receiving the first audio signal and for providing a second filtered signal, a comparator for receiving the first filtered signal and the second filtered signal and for performing a comparison of the first filtered with the second filtered signal and for providing a comparison result to thereby provide the first audio signature signal.

In yet another embodiment of the apparatus for providing an indication of a lip sync, the first video signature extraction unit comprises a first image pixel data providing unit for receiving the first video signal and for providing a corresponding pixel data of a pixel k of a plurality of given pixels 1 to N of a first image of the first video signal; a second image pixel data providing unit for receiving the first video signal and for providing a corresponding pixel data Cic of the corresponding given pixel k in a second image of the first video signal; a comparator for receiving the corresponding pixel data and the corresponding pixel data and for comparing the corresponding pixel data with the corresponding pixel data to provide a result signal indicative of the comparison; a counter for receiving and counting the result signal of each of the plurality of given pixels 1 to N and for providing a counter signal; and a video signature providing unit for receiving the counter signal and for providing the first video signature signal.

In yet another embodiment of the apparatus for providing an indication of a lip sync, the second audio signature extraction unit comprises a first filtering unit for receiving the second audio signal and for providing a first filtered signal; a second filtering unit for receiving the second audio signal and for providing a second filtered signal; a comparator for receiving the first filtered signal and the second filtered signal and for performing a comparison of the first filtered with the second filtered signal and for providing a comparison result to thereby provide the second audio signature signal.

In yet another embodiment of the apparatus for providing an indication of a lip sync, the second video signature extraction unit comprises a first image pixel data providing unit for receiving the second video signal and for providing a corresponding pixel data] of a pixel k of a plurality of given pixels 1 to N of a first image of the second video signal; a second image pixel data providing unit for receiving the second video signal and for providing a corresponding pixel data C1 of the corresponding given pixel k in a second image of the second video signal; a comparator for receiving the corresponding pixel data 1' and the corresponding pixel data 7k and for comparing the corresponding pixel data k with the corresponding pixel data C, to provide a result signal indicative of the comparison; a counter for receiving and counting the result signal of each of the plurality of given pixels 1 to N and for providing a counter signal and a video signature providing unit for receiving the counter signal and for providing the second video signature signal.

Another advantage of the embodiment disclosed is that it may be used for content comparison instead of lip sync detection.

Alternatively, the embodiment disclosed may be used for detecting pirated copies or for checking that a given ads has been properly inserted in a video signal. This may be done by comparing the video signal carrying the ads and the ads per se.

BRIEF DESCRIPTION OF THE DRAWINGS

In order that the invention may be readily understood, embodiments of the invention are illustrated by way of example in the accompanying drawings.

Figure 1 is a flowchart which shows an embodiment of a method for generating a signature representative of a content of a video signal; Figure 2 is a flowchart which shows an embodiment of a method for generating a signature representative of a content of an audio signal; Figure 3A is a block diagram which shows an embodiment of an apparatus for generating a video signature representative of a content of a video signal; Figure 3B is a schematic which shows how the video signature representative of a content of a video signal is generated according to one embodiment; Figure 3C is a block diagram which shows an embodiment of a video signature extraction unit; Figure 4A is a block diagram which shows an embodiment of an apparatus for generating an audio signature representative of a content of an audio signal; Figure 4B is a block diagram which shows an embodiment of an apparatus for generating an audio signature representative of a content of an audio signal; in this embodiment, the apparatus for generating an audio signature representative of a content of an audio signal comprises, inter al/a, an envelope detector, a mean detector and a comparator; Figure 5 is a block diagram of an embodiment of an apparatus comprising an audio signal signature extraction unit, a video signal signature extraction unit and a signature analysis unit for providing, inter a/ia, an indication of a lip sync; Figure 6 is a block diagram which shows an embodiment of the signature analysis unit disclosed in Fig. 5; Figure 7A is a schematic which shows an embodiment of two streams of video signature signals delayed in time; Figure 7B is a block diagram which shows an embodiment of a video signature analysis unit used for analyzing two video signature signals; Figure 7C is a graph which shows an embodiment of a variation of the result of the convolution of a first video signature signal with a second video signature signal; Figure BA is a schematic which shows an embodiment of two streams of audio signature signals delayed in time; Figure BB is a block diagram which shows an embodiment of an audio signature analysis unit used for analyzing two audio signature signals; and Figure BC is a graph which shows an embodiment of a variation of the result of the convolution of a first audio signature signal with a second audio signature signal.

Further details of the invention and its advantages will be apparent from the

detailed description included below.

DETAILED DESCRIPTION

In the following description of the embodiments, references to the accompanying drawings are by way of illustration of an example by which the invention may be practiced. It will be understood that other embodiments may be made without departing from the scope of the invention disclosed.

Now referring to Fig. 1, there is shown an embodiment of a method for generating a video signature representative of a content of a video signal. It will be appreciated that generating a video signature representative of a content of a video signal may be of great advantage for various reasons as further explained herein below.

According to processing step 102, pixel data of each of a plurality of given pixels in a first image are received.

It will be appreciated that in one embodiment, the image originates from a digital video stream. Alternatively, the image may originate from an analog video stream. In other embodiments, the image may originate from a file-based source.

The skilled addressee will appreciate that the given pixels are pixels that may be selected in the image according to various criteria. For instance, the given pixels may be selected according to spatial criteria. For instance, it has been contemplated that some parts of an image may be of less interest for the purpose of generating a signature.

This may be the case for instance for parts of an image where no significant changes occur. Alternatively, it may be the case for parts of an image resulting from a format conversion. Alternatively, those parts of an image may be parts of images that have been modified for branding reasons.

In one embodiment a windowing may be accordingly performed on a part of interest of an image in order to remove those parts of the image that have a limited interest.

In a preferred embodiment, the given pixels are pixels each separated from one another by a given amount of pixels. The amount of given pixels does not change when the resolution of the video signal changes in the preferred embodiment.

Alternatively, the amount of given pixels changes when the resolution of the video signal changes.

According to processing step 104, pixel data of each of a plurality of given pixels in a second image are received.

It will be appreciated that the second image is an image following in time the first image in the video signal.

In a preferred embodiment, and in the case of progressive scan, the second image is not the image following immediately after the first image in the video signal but the image following immediately after the image following immediately after the first image in the video signal.

In a preferred embodiment, and in the case of interlaced scan, a first field of a first image is compared with a first field of a second image immediately following the first image, while a second field of the first image is compared with a second field of the second image immediately following the first image.

According to processing step 106, a comparison is performed.

In one embodiment, the comparison is performed for each given pixel of the plurality of given pixels.

More precisely, corresponding pixel data of the given pixel of the first image is compared with the corresponding pixel data of the given pixel in the second image in order to provide a corresponding indication of a difference between the pixel data of the given pixels of the first image and the pixel data of the given pixel in the second image.

In one embodiment, the comparison is a subtraction of the pixel data of the given pixel of the first image with corresponding pixel data of the given pixel of the second image. The comparison is performed for each given pixel of the plurality of given pixels.

Alternatively the comparison may be a combination of operations involving the pixel data of each of the given pixels of the first image with corresponding pixel data of each of the given pixels of the second image.

According to processing step 108, a counter value is incremented based on the result of the comparison.

In one embodiment, the counter value is incremented in the case where the result of the operation is greater than a given threshold value.

The skilled addressee will appreciate that various embodiments of the given threshold value may be provided. In a preferred embodiment, the threshold value is equal to thirty two (32) on an 8 bit precision video pixel.

Moreover, it will be appreciated that the given threshold value may be provided according to various criteria such as a type of video signal.

According to processing step 110, an indication of the counter value is provided.

The indication of the counter value is used as the video signature representative of a content of the video signal.

It will be appreciated that in one embodiment, the processing step of providing the indication of a counter value may comprise normalizing the counter value to provide a counter value limited by a given value.

In a preferred embodiment, a normalizing of the counter value is performed by dividing the counter value by (N/240) where N is the number of given pixels.

The skilled addressee will appreciate that the method disclosed for providing a video signature representative of a content of the video signal is of great advantage for various reasons.

Another advantage of the method disclosed is that it requires limited processing resources for generating the signature which is therefore of great advantages.

Such method does not require any complex algorithm for analyzing the content of the video signal.

Another advantage of the method is that its implementation requires limited storing resources.

Now referring to Fig. 2, there is shown an embodiment of a method for generating an audio signature representative of a content of an audio signal. It will be appreciated that generating an audio signature representative of a content of an audio signal may be of great advantage for various reasons as further explained below.

According to processing step 202, an audio signal is received. It will be appreciated that the audio signal may be received from various sources. For instance, the audio signal may be receiving from an audio stream. Alternatively, the audio signal may be received from a file. It will further be appreciated that the audio signal may be imbedded in a video stream.

Moreover, it will be appreciated that the audio signal may be provided in various forms such as in a digital format or in an analog format. Moreover it will be appreciated that the audio signal may be formatted according to various standards known to the skilled addressee.

According to processing step 204, a first filtering of the received audio signal is performed.

In one embodiment, the first filtering of the received audio signal comprises detecting an envelope of the audio signal.

According to processing step 206, a second filtering of the received audio signal is performed.

In one embodiment, the second filtering of the received audio signal comprises computing an average value of the received audio signal.

It will be appreciated by the skilled addressee that processing steps 204 and 206 may be performed in parallel. Alternatively, processing steps 204 and 206 may be performed serially.

According to processing step 208, the first filtered audio signal is compared to the second filtered audio signal.

It will be appreciated that the comparison may be a combination of operations involving the first filtered audio signal and the second filtered audio signal.

In a preferred embodiment, the comparison comprises checking if the first filtered audio signal is greater than the second filtered audio signal.

According to processing step 210, a value is assigned depending on the result of the comparison. It will be appreciated that the value may be any type of value.

In a preferred embodiment, the value is a binary value. Still in a preferred embodiment, binary value one (1) is assigned if the first filtered audio signal is greater than the second filtered audio signal while binary value zero (0) is assigned if the second filtered audio signal is greater or equal than the first filtered audio signal.

According to processing step 212, the assigned value is provided. It will be appreciated that the assigned value. The assigned value is used as the audio signature representative of a content of the audio signal.

It will be appreciated that in a preferred embodiment the providing of the assigned value may comprise a decimation processing step. The skilled addressee will appreciate that the purpose of the decimation processing step is to remove a given amount of unwanted/redundant data. The skilled addressee will also appreciate that this will further result in a signature having a shorter size which is also of great advantage.

The skilled addressee will appreciate that the method disclosed for providing an audio signature representative of a content of the audio signal is of great advantage for various reasons.

Another advantage of the method disclosed above is that they require limited processing resources for generating the signature which is of great advantage.

A further advantage is that the implementation of the method disclosed may require limited memory resource which is also of great advantage.

Now referring to Fig. 3a, there is shown an embodiment of an apparatus 300 for providing a video signature.

The apparatus 300 for providing a video signature comprises an optional windowing unit 302, a filtering unit 303 and a video signature extraction unit 304.

More precisely, the optional windowing unit 302 is used for performing a windowing of an incoming image signal í.ç. The optional windowing unit 302 provides a corresponding selected signal w. The skilled addressee will appreciate that the optional windowing unit 302 may be implemented according to various embodiments known to the skilled addressee.

The apparatus 300 for providing a video signature further comprises the filtering unit 303. The filtering unit 303 is used for filtering the corresponding selected signal W. The filtering unit 303 provides a filtered signal íç. It will be appreciated that the filtering unit 303 operates according to various embodiments.

In the case of lOSOiIp and according to a preferred embodiment, the filtering is performed as follows: a filtered pixel data of a given pixel is equal to the average of the pixel data of the given pixel, the pixel immediately following the given pixel on a same line and the pixel immediately preceding the given pixel on the same line.

In the case of 720p and according to a preferred embodiment, the filtering is performed as follows: a filtered pixel data of a given pixel is equal to the average of the pixel data of the given pixel and the pixel immediately following the given pixel on a same line.

In the case of SD and according to a preferred embodiment, no filtering is performed.

The apparatus 300 for providing a video signature further comprises the video signature extraction unit 304.

The video signature extraction unit 304 is used for extracting a video signature from an incoming video signal. In this embodiment, the video signature extraction unit 304 receives the filtered signal provided by the filtering unit 303 and provides a video signature signal VS.

Now referring to Fig. 3B, there is shown an embodiment which shows how the video signature extraction unit 304 operates according to an embodiment.

In accordance with this embodiment, a corresponding pixel data 1k of a pixel k of a plurality of given pixels 1 to N of a first image is compared with a corresponding pixel data k of the corresponding given pixel k in a second image in order to provide a corresponding indication of a difference I-. -C between the pixel data of the given pixel of the first image and the pixel data of the given pixel in the second image.

An absolute value of the result, T -Cj, of the comparison is taken. In the case where the absolute value of the result is larger than a given threshold value, binary value one (1) is assigned. If this is not the case, binary value zero (0) is assigned. Such operation is performed for each pixel of the plurality of given pixels.

Moreover, an optional normalization of the result may be performed. The video signature signal may be therefore defined as: 0 -N/240 The skilled addressee will appreciate that various other alternative embodiments may be provided.

Now referring to Fig. 3C, there is shown an embodiment of the video signature extraction unit 304.

The video signature extraction unit 304 comprises a first image pixel data providing unit 306, a second image pixel data providing unit 308, a comparator 310, a counter 312 and a video signature providing unit 314.

The first image pixel data providing unit 306 is used for receiving the filtered signal W provided by the filtering unit 303 and for providing a corresponding pixel data F of a pixel k of a plurality of given pixels 1 to N of the first image.

Similarly, the second image pixel data providing unit 308 is used for receiving the filtered signal W1 provided by the filtering unit 303 and for providing a corresponding pixel data k of the corresponding given pixel k in the second image.

The comparator 310 is used for receiving the corresponding pixel data F of a pixel k of a plurality of given pixels 1 to N of the first image and the corresponding pixel data C, of the corresponding given pixel k in the second image and for comparing the corresponding pixel data 1 with the corresponding pixel data C1 In one embodiment, the comparator 310 outputs a logic value one (1) if 4 -Ck is greater than thirty two (32) and a logic value zero (0) otherwise.

The counter 312 receives the output from the comparator 310 and provides a signal indicative of a number of logic value ones received by the counter 312.

The video signature providing unit 314 receives the signal indicative of a number of logic value ones received by the counter 312 and provides the video signature signal TCS. In one embodiment, the video signature providing unit 314 performs a division of the signal indicative of a number of logic value ones received by the counter 312 by N/240 wherein N is the number of given pixels.

Now referring to Fig. 4A, there is shown an embodiment of an apparatus 402 for generating an audio signature representative of a content of an audio signal.

In this embodiment, the apparatus 402 for generating an audio signature representative of a content of an audio signal receives an audio signal 4 and provides a corresponding audio signature signal AS.

It will be appreciated that the audio signal A, may be of various forms such as in a digital format or in an analog format. Moreover it will be appreciated that the audio signal may be formatted according to various standards as already explained above.

Similarly, it will be appreciated that the corresponding audio signature signal AS may be of various forms such as in a digital format or in an analog format.

Moreover it will be appreciated that the corresponding audio signature signal AS1 may be formatted according to various standards as explained above.

Now referring to Fig. 4B, there is shown an embodiment of the apparatus 402 for generating an audio signature representative of a content of an audio signal.

In this embodiment, the apparatus 402 for generating an audio signature representative of a content of an audio signal comprises an absolute value providing unit 404, an envelope detector 406, a mean detector 408, a comparator 410 and a decimator 412.

The absolute value providing unit 404 is used to provide a signal indicative of an absolute value of the audio signal A,.

The envelope detector 406 is used to provide a signal E indicative of an envelope of the signal indicative of an absolute value of the value of the audio signal 4. It will be appreciated that the envelope detector 406 is an embodiment of a first filtering unit.

Moreover, it will be appreciated that the envelope detector 406 may be implemented in various ways as known by the skilled addressee. -21 -

In a preferred embodiment, the envelope detector 406 is implemented using a one-tap Infinite Impulse Response (IIR) filter.

The mean detector 408 is used to provide a signal M indicative of a mean of the signal indicative of an absolute value of the value of the audio signal A,. It will be appreciated that the mean detector 408 is an embodiment of a second filtering unit.

Moreover, it will be appreciated that the mean detector 408 may be implemented in various ways as known by the skilled addressee.

In a preferred embodiment, the mean detector 408 is implemented using a one-tap Infinite Impulse Response (IIR) filter.

The comparator 410 is used for making a comparison between two incoming signals. More precisely, the comparator 410 receives the signal M indicative of a mean of the signal indicative of an absolute value of the value of the audio signal A, and the signal E indicative of an envelope of the signal indicative of an absolute value of the value of the audio signal A, and performs a comparison between those two signals.

If the signal E indicative of an envelope of the signal indicative of an absolute value of the value of the audio signal A, is greater than the signal M5 indicative of a mean of the signal indicative of an absolute value of the value of the audio signal 4, a one (1) logic value signal is outputted by the comparator 410.

Otherwise a zero (0) logic value signal is outputted by the comparator 410.

The skilled addressee will appreciate that various other types of comparison may be made.

Moreover, it will be appreciated that the comparator 410 may be implemented in various ways.

The decimator 412 is used for performing decimation on a signal provided by the comparator 410. The decimator 412 provides the audio signature signal AS.

It will be appreciated that the decimator 412 is optional.

Moreover, it will be appreciated that the decimator 412 may be implemented in various ways as known by the skilled addressee. In one embodiment, a decimation by fifty two (52) is performed by the decimator 412.

In a preferred embodiment, the decimator 412 takes a sample and ignores fifty one (51) following samples. Alternatively, the sample is taken every one (1) ms.

Other methods may alternatively be used. For instance, those methods may take into consideration the value of each sample for selecting a given sample.

Now referring to Fig. 5, there is shown an embodiment of an apparatus 500 which uses an audio signal signature extraction unit and a video signal signature extraction unit for providing, inter a/ia, an indication of a lip sync.

In this embodiment, a first audio/video content is to be compared to a second audio/video content.

More precisely, the first audio/video content comprises a first audio signal and a first video signal while the second audio/video content comprises a second audio signal an a second video signal.

As shown below, by carefully comparing the signatures of the first audio/video content with the signatures of the second audio/video content it is possible to provide an indication of a lip sync, i.e. to providing an indication of a desynchronization that occurred between the video signal and its corresponding audio signal. As known by the skilled addressee, such desynchronization may occur for various reasons such as for instance for reasons associated with the transmission of the signals over a medium.

More precisely and as shown in Fig. 5, the apparatus 500 comprises a first signature extraction unit 502, a second signature extraction unit 504 and a signature analysis unit 506.

The first signature extraction unit 502 is used for providing a first video signature signal VS and a first audio signature signal AS1 of respectively a first video signal T and a first audio signal A1 associated with the first video signal v.

The first signature extraction unit 502 comprises, as not shown in Fig. 5, an audio signature extraction unit and a video signature extraction unit each responsible for respectively receiving the first audio signal A1 and providing the first audio signature signal AS1 and receiving the first video signal i and providing the first video signature signal VS. Such video signature extraction unit and audio signature extraction unit have been already described above.

The second signature extraction unit 504 is used for providing a second video signature signal VS., and a second audio signature signal AS., of respectively a second video signal J' and a second audio signal A, associated with the second video signal r.

The second signature extraction unit 504 comprises, as not shown in Fig. 5, an audio signature extraction unit and a video signature extraction unit each responsible for respectively receiving the second audio signal A., and providing the second audio signature signal AS, and receiving the second video signal PT, and providing the second video signature signal VS.,. Such video signature extraction unit and audio signature extraction unit have also been already described above.

The signature analysis unit 506 is used for receiving the first video signature signal J"S, the first audio signature signal AS1, the second video signature signal PS2 and the second audio signature signal AS2.

The signature analysis unit 506 provides a signal VD indicative of a video delay, a signal AD indicative of an audio delay, a signal LS indicative of a lip sync, a signal VMF indicative of a video matching factor and a signal AA'If' indicative of an audio matching factor.

The signal VD indicative of a video delay is generated by comparing the first video signature signal V5 with the second video signature signal VS'2 and by determining a delay between those two signals as shown further below.

Similarly, the signal AD indicative of an audio delay is generated by comparing the first audio signature signal AS1 with the second audio signature signal AS, and by determining a delay between those two signals as shown further below.

The signal JMF indicative of a video matching factor and the signal AMF indicative of an audio matching factor are generated as further explained below.

Finally and in one embodiment, the signal LS indicative of a lip sync is generated by comparing the signal VD indicative of a video delay with the signal AD indicative of an audio delay.

In the implementation described herein the signal J3\//F indicative of a video matching factor and the signal AJ%IF indicative of an audio matching factor are further used for generating the signal LS indicative of a lip sync.

While the embodiment disclosed in Fig. 5 has been shown to be used for lip sync detection, the skilled addressee will appreciate that such embodiment may be advantageously used for content comparison for instance.

Alternatively, the embodiment may be used for detecting pirated copies. In an alternative embodiment, the method disclosed may be used for checking that a given ads has been properly inserted in a video signal. This may be done by comparing the signatures of the audio/video signal carrying the ads and the signatures of the ads per se.

Now referring to Fig. 6, there is shown an embodiment of the signature analysis unit 506 disclosed in Fig. 5.

The signature analysis unit 506 comprises a video signature analysis unit 602, an audio signature analysis unit 604 and a video signature correlation analysis unit 606.

The video signature analysis unit 602 is used for determining an estimation VD' of the signal indicative of a video delay between the first video signature signal VS1 and the second video signature signal VS2.

The video signature analysis unit 602 is further used for determining an estimation T'ME' of a signal indicative of a video matching factor.

Similarly, the audio signature analysis unit 604 is used for determining an estimation Al)' of the signal indicative of an audio delay between the first audio signature signal AS1 and the second audio signature signal AS2.

The audio signature analysis unit 604 is further used for determining an estimation AMP' of a signal indicative of an audio matching factor.

The video signature correlation analysis unit 606 receives the estimation kD' of the signal indicative of a video delay between the first video signature signal VS1 and the second video signature signal frIS'2, the estimation AD' of the signal indicative of an audio delay between the first audio signature signal AAS and the second audio signature signal AS2, the estimation FMF' of a signal indicative of a video matching factor and the estimation A/VP' of a signal indicative of an audio matching factor.

The video signature correlation analysis unit 606 provides the signal LS indicative of a lip sync, the signal VD indicative of a video delay, the signal Al) indicative of an audio delay, the signal VAlE indicative of a video matching factor and the signal AME indicative of an audio matching factor.

In one embodiment, the signal LX indicative of a lip sync is defined as LX = (VD'-ATY) f VMP»= threshold] & AMP»= threshoid2, wherein thresholdi and threshold2 are given values. In one embodiment, the thresholdi is equal to 50% while the threshold2 is equal to 40%. The skilled addressee will appreciate that various other values may be used.

The skilled addressee will appreciate that the signal VD indicative of a video delay and the signal AD indicative of an audio delay are generated using respectively at least the estimation VD' of the signal indicative of a video delay between the first video signature signal VS1 and the second video signature signal VS2 and the estimation AD' of the signal indicative of an audio delay between the first audio signature signal AS and the second audio signature signal AS2. In fact, the skilled addressee will appreciate that estimations are first computed prior to provide the values for the sake validating the values.

Now referring to Fig. 7A, there are shown a graph 702 showing an example of the first video signature signal J75 and a graph 704 showing an example of the second video signature signal V2.

It will be appreciated that the first video signature signal V1SI and the second video signature signal VS2 are desynchronized in time by an amount of time corresponding to the estimation FLY of the signal indicative of a video delay between the first video signature signal V and the second video signature signal VS2. Still in this embodiment, the first video signature signal VS1 is delayed in time compared to the second video signature signal VS2.

Now referring to Fig. 7B, there is shown an embodiment of the video signature analysis unit 602.

The video signature analysis unit 602 comprises a convolution unit 706 and a minimum detection and delay extraction unit 708.

The convolution unit 706 is used for performing a convolution of the first video signature signal VS1 and the second video signature signal VS2. In one embodiment the convolution is a time shifted convolution of the second video signature signal frS window across the first video signature signal k1S window.

In one embodiment, the second video signature signal VS, window has a value of twenty (20) seconds while the first video signature signal V5 window has a value of thirty (30) seconds.

The convolution unit 706 provides a convolution signal E(t).

It will be appreciated by the skilled addressee that the convolution may be alternatively performed in various other ways. Moreover it will be appreciated that the convolution unit 706 may implemented in various way.

The minimum detection and delay extraction unit 708 is used for detecting a minimum in the convolution signal /?(t). It will be appreciated that the minimum in the convolution signal Lc,(/) is indicative of the estimation VII of the signal indicative of a video delay between the first video signature signal VS1 and the second video signature signal VS2.

The minimum detection and delay extraction unit 708 is further used for providing the estimation kMF' of a signal indicative of a video matching factor.

It will be appreciated that the signal P7<IF indicative of a video matching factor is indicative of a level of similarity measured in the video signatures.

It will be appreciated that the signal estimation V?VJF' of a signal indicative of a video matching factor is a function of the convolution signal E (I) and the estimation VI)' of the signal indicative of a video delay between the first video signature signal VS1 and the second video signature signal VS.,.

In a preferred embodiment, the signal estimation J'2t1F' of a signal indicative of a video matching factor is defined as VMF=IOO*(1_(M/n(EJi))/Surn(V52)), wherein kAlE' is the estimated video matching factor between the first video signature signal VS1 and the second video signature signal VS2 and is expressed in percentage, A'Iin(E1.(t)) is the minimum error found in the video correlation graph disclosed at Fig. 7c., and Su;n(VS.,) is the sum of the signature vector in the window W.,. It will be appreciated that in that embodiment, when V and V, are exactly the same, J'MF' is equal to 100% while if one of the two video sources is altered, PAlE' will be reduced. In this embodiment, under 50%, the video sources are considered to be different.

It will be appreciated that the minimum detection and delay extraction unit 708 may be implemented in various ways.

Now referring to Fig. 8, there are shown a graph 800 showing an example of the first audio signature signal AS1 and a graph 802 showing an example of the second audio signature signal AS2.

It will be appreciated that the first audio signature signal AS1 and the second audio signature signal AS, are desynchronized in time by an amount of time corresponding to the estimation AD' of the signal indicative of an audio delay between the first audio signature signal AS1 and the second audio signature signal AS2. Still in this embodiment, the first audio signature signal AS1 is delayed in time compared to the second audio signature signal AS2.

Now referring to Fig. SB, there is shown an embodiment of audio signature analysis unit 604 for determining the estimation AD' of the signal indicative of an audio delay between the first audio signature signal AS1 and the second audio signature signal AS, and the estimation A/vIE' of a signal indicative of an audio matching factor.

The audio signature analysis unit 604 comprises a convolution unit 804 and a minimum detection and delay extraction unit 806.

The convolution unit 804 is used for performing a convolution of the first audio signature signal AS1 and the second audio signature signal AS2. In one embodiment the convolution is a time shifted convolution of the second audio signature signal AS7 window across the first audio signature signal AS1 window.

In one embodiment, the second audio signature signal AS2 window has a value of one (1) second while the first audio signature signal AS1 window has a value of ten (10) seconds. It will be appreciated that those values may be changed depending on various needs.

The convolution unit 804 provides a convolution signal EQ).

It will be appreciated by the skilled addressee that the convolution may be alternatively performed in various other ways. Moreover it will be appreciated that the convolution unit 804 may implemented in various ways.

The minimum detection and delay extraction unit 806 is used for detecting a minimum in the convolution signal.E4(t). It will be appreciated that the minimum in the convolution signal E4Q) is indicative of the estimation Al)' of the signal indicative of an audio delay between the first audio signature signal AAS and the second audio signature signal AS2.

The minimum detection and delay extraction unit 806 is further used for providing the signal estimation AMP of a signal indicative of an audio matching factor.

It will be appreciated that the signal estimation AMP of a signal indicative of an audio matching factor is indicative of a level of similarity measured in the audio signatures.

It will be appreciated that the signal estimation AMP of a signal indicative of an audio matching factor is a function of the convolution signal E4Q) and the estimation AD' of the signal indicative of an audio delay between the first audio signature signal AS1 and the second audio signature signal AS2.

In a preferred embodiment, the signal estimation A/VIE' of a signal indicative of an audio matching factor is defined as: AMF=100*(1_M/n((E4(t))/(S/ze(AS,)/5)), wherein AIvIF' is the estimated audio matching factor between the first audio signature signal AS1 and the second audio signature signal AS, expressed in percentage, MEnUF4(r)) is the minimum error found in the audio correlation graph shown in Fig. Sc and Size(A52) is the size of the audio signature vector in the window W. It will be appreciated that when A1 and A, are exactly the same, the signal estimation AIvIF' of a signal indicative of an audio matching factor is equal to 100%. If one of the two audio sources was altered, the signal estimation A/VIE' of a signal indicative of an audio matching factor will be reduced. Still in this embodiment, when the signal estimation Ak/F' of a signal indicative of an audio matching factor is under 40%, the audio sources are considered to be different.

It will be appreciated that the minimum detection and delay extraction unit 806 may be implemented in various ways. For instance, the minimum detection and delay extraction unit 806 may first perform a pre-scan in order to select a first selection of convoluted samples. In one embodiment the first selection represents one-eight (1/8) of all the convoluted samples. A search may then be performed in the selected convoluted samples.

Although the above description relates to a specific preferred embodiment as presently contemplated by the inventor, it will be understood that the invention in its broad aspect includes mechanical and functional equivalents of the elements described herein.

Claims

CLAIMS: 1. A method for generating an audio signature representative of a content of an audio signal, the method comprising: receiving an audio signal; performing a first filtering of the audio signal to provide a first filtered audio signal; performing a second filtering of the audio signal to provide a second filtered audio signal; comparing the first filtered audio signal to the second filtered audio signal; and assigning a value depending on the comparison to thereby generate the audio signature representative of the content of the audio signal.
2. The method as claimed in claim 1, wherein said first filtering comprises detecting an envelope signal of the received audio signal.
3. The method as claimed in any one of claims 1 to 2, wherein said second filtering comprises computing an average value of the received audio signal.
4. The method as claimed in claim 3, wherein the assigning of a value depending on the comparison to thereby generate the audio signature representative of the content of the audio signal comprises performing a decimation.
5. An audio signature extraction unit for providing an audio signature signal AS, representative of a content of an audio signal4, the audio signature extraction unit comprising: a first filtering unit for receiving the audio signal A, and for providing a first filtered signal; a second filtering unit for receiving the audio signal A and providing a second filtered signal; a comparator for receiving the first filtered signal and the second filtered signal and for performing a comparison of the first filtered with the second filtered signal and for providing a comparison result to thereby provide the audio signature signal AS, representative of a content of an audio signal.
6. The audio signature extraction unit as claimed in claim 5, wherein said first filtering unit comprises an envelope detector for providing a signal E indicative of an envelope of the audio signalA1.
7. The audio signature extraction unit as claimed in any one of claims 5 to 6, wherein said second filtering unit comprises a mean detector for providing a signal Ai indicative of a mean the audio signal A,.
8. The audio signature extraction unit as claimed in claim 7, further comprising an absolute value providing unit for receiving the audio signal A1 and for providing a signal indicative of an absolute value of the value of the audio signal 4; wherein each of said mean detector and said envelope detector receives said signal indicative of an absolute value of the value of the audio signal A1.
9. The audio signature extraction unit as claimed in any one of claims 5 to 8, further comprising a decimator for receiving the audio signature signal A5 and for decimating the audio signature signal AS,.
10. An apparatus for providing an indication of a lip sync, the apparatus comprising: a first signature extraction unit for receiving a first audio signal and a first video signal and for providing a first video signature signal and a first audio signature signal; a second signature extraction unit for receiving a second audio signal and a second video signal and for providing a second video signature signal and a second audio signature signal; a signature analysis unit for receiving the first video signature signal, the first audio signature signal, the second video signature signal and the second audio signature signal and for correlating the first video signature signal with the second video signature signal to provide a video delay and for further correlating the first audio signature signal with the second audio signature signal to provide an audio delay, said signature analysis unit for further comparing said audio delay with said video delay and for providing an indication of a lip sync is said audio delay is different from said video delay.
11. The apparatus as claimed in claim 10, wherein said first signature extraction unit comprises a first audio signature extraction unit for receiving said first audio signal and for providing said first audio signature signal, further wherein said first signature extraction unit comprises a first video signature extraction unit for receiving said first video signal and for providing said first video signature signal.
12. The apparatus as claimed in any one of claims 10 to 11, wherein said second signature extraction unit comprises a second audio signature extraction unit for receiving said second audio signal and for providing said second audio signature signal, further wherein said second signature extraction unit comprises a second video signature unit for receiving said second video signal and for providing said second video signature signal.
13. The apparatus as claimed in claim 11, wherein said first audio signature extraction unit comprises: a first filtering unit for receiving the first audio signal and for providing a first filtered signal; a second filtering unit for receiving the first audio signal and for providing a second filtered signal; a comparator for receiving the first filtered signal and the second filtered signal and for performing a comparison of the first filtered with the second filtered signal and for providing a comparison result to thereby provide the first audio signature signal.
14. The apparatus as claimed in any one of claims 11 and 13, wherein said first video signature extraction unit comprises: a first image pixel data providing unit for receiving said first video signal and for providing a corresponding pixel data of a pixel k of a plurality of given pixels 1 to N of a first image of said first video signal; a second image pixel data providing unit for receiving said first video signal and for providing a corresponding pixel data of the corresponding given pixel k in a second image of said first video signal; a comparator for receiving the corresponding pixel data and the corresponding pixel data C, and for comparing the corresponding pixel data with the corresponding pixel data to provide a result signal indicative of the comparison; a counter for receiving and counting the result signal of each of the plurality of given pixels 1 to N and for providing a counter signal; and a video signature providing unit for receiving the counter signal and for providing the first video signature signal.
15. The apparatus as claimed in claim 12, wherein said second audio signature extraction unit comprises: a first filtering unit for receiving the second audio signal and for providing a first filtered signal; a second filtering unit for receiving the second audio signal and for providing a second filtered signal; a comparator for receiving the first filtered signal and the second filtered signal and for performing a comparison of the first filtered with the second filtered signal and for providing a comparison result to thereby provide the second audio signature signal.
16. The apparatus as claimed in any one of claims 12 and 15, wherein said second video signature extraction unit comprises: a first image pixel data providing unit for receiving said second video signal and for providing a corresponding pixel data of a pixel k of a plurality of given pixels 1 to N of a first image of said second video signal; a second image pixel data providing unit for receiving said second video signal and for providing a corresponding pixel data of the corresponding given pixel k in a second image of said second video signal; a comparator for receiving the corresponding pixel data and the corresponding pixel data and for comparing the corresponding pixel data with the corresponding pixel data to provide a result signal indicative of the comparison; a counter for receiving and counting the result signal of each of the plurality of given pixels 1 to N and for providing a counter signal; and a video signature providing unit for receiving the counter signal and for providing the second video signature signal.