US 10,475,462 B2Granted

Audio recognition apparatus and method

AssigneePlayfusion Limited

InventorsRiaan Hodgson, David Gomberg, Mark Gerhard

Life of this patent

Estimating the term…

Loading patent…

Abstract

A method includes generating, by a processor, an audio fingerprint representative of an audio signal. The audio fingerprint is based on a plurality of first intensity values corresponding to one or more segments of the audio signal. The plurality of first intensity values are based on a Fast Fourier Transform (FFT) performed on at least one sampled segment of the audio signal. The method also includes comparing a plurality of second intensity values based on a recorded sound to determine whether the second intensity values match the first intensity values. The method additionally includes causing a message to be communicated to a device used to record the sound based on a determination that the plurality of second intensity values match the plurality of first intensity values.

Claims16

Dependent claims

A method, comprising:

generating, by a processor, an audio fingerprint representative of an audio signal, the audio fingerprint being based on a plurality of first intensity values corresponding to one or more segments of the audio signal, the plurality of first intensity values being based on a Fast Fourier Transform (FFT) performed on at least one sampled segment of the audio signal;
comparing a plurality of second intensity values based on a recorded sound to determine whether the second intensity values match the first intensity values;
and
causing a message to be communicated to a device used to record the sound based on a determination that the plurality of second intensity values match the plurality of first intensity values,
wherein
generating the audio fingerprint further comprises performing the FFT on a plurality of overlapped frames of the audio signal,
the second intensity values of the plurality of second intensity values are generated free from a calculation involving overlapped frames, and
the device used to record the sound is remote from the processor, and the comparison is performed by the processor.

5 more elements

A method, comprising:

sampling an audio signal at a sampling rate;
segmenting the sampled audio signal into at least a first frame having a first quantity of samples and a second frame having a second quantity of samples;
generating a plurality of first intensity values by performing a first Fast Fourier Transform (FFT) on the samples included in the first frame;
generating a plurality of second intensity values by performing a second FFT on the samples included in the second frame;
mixing the plurality of first intensity values and the plurality of second intensity values to generate a plurality of average intensity values;
dividing a predetermined audio frequency range into a set of frequency bands, wherein each frequency band of the set of frequency bands has a low end and a high end, and the high end of at least one frequency band of the set of frequency bands is the low end of a next frequency band of the set of frequency bands;
identifying, for each frequency band of the set of frequency bands, a first average intensity value of the plurality of average intensity values closest to the low end of a corresponding frequency band and a second average intensity value of the plurality of average intensity values closest to the high end of the corresponding frequency band;
generating a set of base intensity values comprising a quantity of values equal to a quantity of frequency bands included in the set of frequency bands by averaging the first average intensity value and the second average intensity value corresponding to each frequency band of the set of frequency bands;
generating an audio fingerprint representative of the audio signal based on the set of base intensity values, wherein generating the audio fingerprint comprises performing the first FFT on a plurality of overlapped frames of the audio signal,
comparing a set of third intensity values to the base intensity values to determine if a recorded sound upon which the set of third intensity values is based matches the audio fingerprint;
and
causing a message to be output by the device remote from the computer based on a determination that the recorded sound matches the audio fingerprint,
wherein
the sound is recorded by a device remote from a computer used to determine if the recorded sound matches the audio fingerprint, and
the third intensity values of the plurality of third intensity values are generated free from a calculation involving overlapped frames.

12 more elements

Dependencies are inferred from the claim text.

Facts

Application: 15/806,401Filed 8 Nov 2017
Published: US 2019/0139557 A19 May 2019
Granted: 12 Nov 2019
Term ends: Not estimatedThe source records don’t support an estimate.

Classification4 codes

G10L 19/018
Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals,…:…
G10L 25/51
Specially adapted for particular use: for comparison or discrimination
G10L 25/18
Characterised by the type of extracted parameters: the extracted parameters being spectral information of each sub-band
G10L
Speech analysis techniques or speech synthesis