Audio fingerprint extraction by scaling in time and resampling
View Patent ↗An audio fingerprint is extracted from an audio sample, where the fingerprint contains information that is characteristic of the content in the sample. The fingerprint may be generated by computing an energy spectrum for the audio sample, resampling the energy spectrum, transforming the resampled energy spectrum to produce a series of feature vectors, and computing the fingerprint using differential coding of the feature vectors. The generated fingerprint can be compared to a set of reference fingerprints in a database to identify the original audio content.
1. A method for extracting an audio fingerprint from an audio frame, the method comprising:
filtering, by a processor, the audio frame into a plurality of frequency bands by varying the frequency bands with time to produce a corresponding plurality of filtered audio signals;
scaling in time, by the processor, for fingerprinting, a size of each filtered audio signal based on a frequency of the corresponding frequency band, the scaling based on at least one of a mid-frequency and a frequency range of the corresponding frequency band;
resampling, by the processor, the scaled filtered audio signals by sampling from different corresponding frequency bands as time changes to produce resampled audio signals;
transforming, by the processor, the resampled audio signals to produce a feature vector for each resampled audio signal; and
computing, by the processor, the audio fingerprint based on the feature vectors.
2. The method of claim 1 , wherein transforming a resampled audio signal comprises performing a transform along a time axis.
3. The method of claim 1 , wherein transforming a resampled audio signal comprises:
transforming the resampled audio signal along a time axis; and
transforming the resampled audio signal along a frequency axis.
4. The method of claim 3 , further comprising, by the processor, scaling in magnitude, windowing, and normalizing the resampled audio signal.
5. The method of claim 1 , wherein transforming a resampled audio signal comprises performing a Two Dimensional Discrete Cosine Transform (2D DCT).
6. A computing device, comprising:
a processor;
a storage medium for tangibly storing thereon program logic for execution by the processor, the program logic comprising:
filtering logic executed by the processor for filtering an audio frame into a plurality of frequency bands by varying the frequency bands with time to produce a corresponding plurality of filtered audio signals;
scaling logic executed by the processor for scaling in time, for fingerprinting, a size of each filtered audio signal based on a frequency of the corresponding frequency band, the scaling based on at least one of a mid-frequency and a frequency range of the corresponding frequency band;
resampling logic executed by the processor for resampling the scaled filtered audio signals by sampling from different corresponding frequency bands as time changes to produce resampled audio signals;
transforming logic executed by the processor for transforming the resampled audio signals to produce a feature vector for each resampled audio signal; and
computing logic executed by the processor for computing an audio fingerprint based on the feature vectors.
7. The computing device of claim 6 , wherein the transforming logic for transforming a resampled audio signal comprises performing logic for performing a transform along a time axis.
8. The computing device of claim 6 , wherein the transforming logic for transforming a resampled audio signal comprises:
second transforming logic for transforming the resampled audio signal along a time axis; and
third transforming logic for transforming the resampled audio signal along a frequency axis.
9. The computing device of claim 8 , further comprising second scaling logic executed by the processor for scaling in magnitude, windowing, and normalizing the resampled audio signal.
10. The computing device of claim 6 , wherein the transforming logic for transforming a resampled audio signal comprises performing logic for performing a Two Dimensional Discrete Cosine Transform (2D DCT).
11. A non-transitory computer readable storage medium for tangibly storing computer program instructions capable of being executed by a computer processor, the computer program instructions defining the steps of:
filtering, by the computer processor, an audio frame into a plurality of frequency bands by varying the frequency bands with time to produce a corresponding plurality of filtered audio signals;
scaling, by the computer processor, in time, for fingerprinting, a size of each filtered audio signal based on a frequency of the corresponding frequency band, the scaling based on at least one of a mid-frequency and a frequency range of the corresponding frequency band;
resampling, by the computer processor, the scaled filtered audio signals by sampling from different corresponding frequency bands as time changes to produce resampled audio signals;
transforming, by the computer processor, the resampled audio signals to produce a feature vector for each resampled audio signal; and
computing, by the computer processor, an audio fingerprint based on the feature vectors.
12. The non-transitory computer readable storage medium of claim 11 , wherein transforming a resampled audio signal comprises performing a transform along a time axis.
13. The non-transitory computer readable storage medium of claim 11 , wherein transforming a resampled audio signal comprises:
transforming the resampled audio signal along a time axis; and
transforming the resampled audio signal along a frequency axis.
14. The non-transitory computer readable storage medium of claim 13 , further comprising, by the computer processor, scaling in magnitude, windowing, and normalizing the resampled audio signal.
15. The non-transitory computer readable storage medium of claim 11 , wherein transforming a resampled audio signal comprises performing, by the computer processor, a Two Dimensional Discrete Cosine Transform (2D DCT).