Extraction and matching of characteristic fingerprints from audio signals
View Patent ↗A method, database, and article of manufacture comprising a plurality of audio fingerprints. Each audio fingerprint contains characteristic information about a corresponding audio frame and is produced by filtering the corresponding audio frame into frequency bands, resampling the filtered audio signals at a nonlinear timescale, transforming the resampled audio signals for each frequency band to produce a feature vector for the frequency band, and computing the audio fingerprint based on the set of feature vectors, and one or more index values for one or more of the audio fingerprints, where the audio fingerprints are organized according to their index values.
1. An article of manufacture comprising an audio fingerprint stored on a non-transitory computer readable storage medium, wherein the audio fingerprint contains characteristic information about an audio frame and is produced by a process comprising:
filtering, by a computing device, the audio frame into a plurality of frequency bands to produce a corresponding plurality of filtered audio signals;
resampling, by the computing device, the filtered audio signals at a nonlinear timescale;
transforming, by the computing device, the resampled audio signals for each frequency band to produce a feature vector for the frequency band; and
computing, by the computing device, the audio fingerprint based on the set of feature vectors.
2. The article of manufacture of claim 1 , wherein filtering the audio frame into a plurality of frequency bands comprises band pass filtering the audio frame in each of the plurality of frequency bands.
3. The article of manufacture of claim 1 , wherein filtering the audio frame into a plurality of frequency bands comprises performing a Fast Fourier Transform (FFT) on the audio sample.
4. The article of manufacture of claim 1 , wherein the audio frame is part of an audio file stored in an MP3 format, and the filtered audio signals are obtained from an MP3 hybrid filterbank associated with the audio file.
5. The article of manufacture of claim 1 , wherein the filtered audio signals are resampled at a logarithmic timescale.
6. The article of manufacture of claim 1 , wherein the frequency bands are spaced linearly in a frequency axis.
7. The article of manufacture of claim 1 , wherein the frequency bands overlap.
8. The article of manufacture of claim 1 , wherein transforming the resampled filtered audio signal of a particular frequency band comprises performing a Fast Fourier Transform (FFT) on the resampled audio signal.
9. The article of manufacture of claim 1 , wherein computing the audio fingerprint comprises differentially encoding the feature vectors for the frequency bands.
10. The article of manufacture of claim 1 , wherein the process further comprises: computing an index value for the audio fingerprint, the index value comprising a portion of the audio fingerprint.
11. The article of manufacture of claim 10 , wherein the index value comprises a portion of the audio fingerprint that corresponds to a set of low frequency components of the transformed audio signals.
12. The article of manufacture of claim 1 , wherein the process further comprises:
disregarding a portion of the audio fingerprint, where the disregarded portion of the audio fingerprint corresponds to a frequency range determined to be insignificant according to an acoustic model.
13. The article of manufacture of claim 12 , wherein the acoustic model is a psychoacoustic model.
14. The article of manufacture of claim 12 , wherein the acoustic model mimics the properties of an audio encoding process.
15. The article of manufacture of claim 12 , wherein the acoustic model mimics the properties of an environment.
16. The article of manufacture of claim 12 , wherein the acoustic model mimics the properties of an audio signal.
17. The article of manufacture of claim 1 , wherein transforming the resampled filtered audio signal of a particular frequency band comprises performing a Discrete Cosine Transform (DCT) on the resampled audio signal.
18. The article of manufacture of claim 1 , wherein the frequency bands have a logarithmic mid-frequency distribution in a frequency axis.
19. The article of manufacture of claim 1 , wherein the filtered audio signals are resampled at an exponential timescale.
20. The article of manufacture of claim 1 , wherein computing the audio fingerprint comprises encoding the feature vectors for the frequency bands by assigning bit values according to a codebook table.
21. The article of manufacture of claim 20 , wherein the process further comprises: training the codebook table to determine a set of codebook values; and tuning the codebook table based on the determined set of codebook values.
22. A database of audio fingerprints, the database comprising a non-transitory computer readable storage medium that contains a plurality of audio fingerprints, wherein each audio fingerprint contains characteristic information about a corresponding audio frame and is produced by a process comprising:
filtering, by a computing device, the corresponding audio frame into a plurality of frequency bands to produce a corresponding plurality of filtered audio signals,
resampling, by the computing device, the filtered audio signals at a nonlinear timescale,
transforming, by the computing device, the resampled audio signals for each frequency band to produce a feature vector for the frequency band, and
computing, by the computing device, the audio fingerprint based on the set of feature vectors; and one or more index values for one or more of the audio fingerprints,
wherein the audio fingerprints are organized in the database according to their index values.
23. The database of claim 22 , wherein each index value comprises a portion of an audio fingerprint associated with the index value.
24. An article of manufacture comprising an audio fingerprint stored on a non-transitory computer readable storage medium, wherein the audio fingerprint contains characteristic information about an audio frame and is produced by a process comprising:
computing, by a computing device, a spectrogram for the audio frame;
sampling, by the computing device, the spectrogram at a nonlinear time scale for a plurality of frequency bands in the spectrogram;
extracting, by the computing device, a long-term feature vector using the samples from each of the sampled frequency bands; and
generating, by the computing device, the audio fingerprint based on the feature vectors.
25. The article of manufacture of claim 24 , wherein the spectrogram is sampled at a logarithmic timescale.
26. A method for extracting an audio fingerprint from an audio frame, the method comprising:
filtering, by a computing device, the audio frame into a plurality of frequency bands to produce a corresponding plurality of filtered audio signals;
transforming, by the computing device, the plurality of filtered audio signals for each frequency band using a discrete cosine transform (DCT) to produce a feature vector for the frequency band;
computing, by the computing device, the audio fingerprint based on the set of feature vectors; and
storing, by the computing device, the computed audio fingerprint on a non-transitory computer readable medium.
27. The method of claim 26 , further comprising:
resampling the filtered audio signals at a nonlinear timescale.
28. The method of claim 27 , wherein the filtered audio signals are resampled at a logarithmic timescale.
29. The method of claim 26 , wherein the frequency bands are spaced linearly in a frequency axis.
30. The method of claim 26 , wherein the frequency bands overlap.
31. The method of claim 26 , wherein computing the audio fingerprint comprises differentially encoding the feature vectors for the frequency bands.
32. The method of claim 26 , further comprising:
computing an index value for the audio fingerprint, the index value comprising a portion of the audio fingerprint.
33. The method of claim 32 , wherein the index value comprises a portion of the audio fingerprint that corresponds to a set of low frequency components of the transformed audio signals.
34. The method of claim 26 , further comprising:
disregarding a portion of the audio fingerprint, where the disregarded portion of the audio fingerprint corresponds to a frequency range determined to be insignificant according to an acoustic model.
35. The method of claim 34 , wherein the acoustic model is a psychoacoustic model.
36. The method of claim 34 , wherein the acoustic model mimics the properties of an audio encoded process.
37. The method of claim 34 , wherein the acoustic model mimics the properties of an environment.
38. The method of claim 34 , wherein the acoustic model mimics the properties of an audio signal.