Method and apparatus for speech recognition, and electronic device

Patent metadata

Field	Value
Publication number	US-11217229-B2
Application number	US-202016921537-A
Country	US
Kind code	B2
Filing date	Jul 6, 2020
Priority date	Jun 28, 2018
Publication date	Jan 4, 2022
Grant date	Jan 4, 2022

Abstract

A speech recognition method, apparatus, a computer device and an electronic device for recognizing speech. The method includes receiving an audio signal obtained by a microphone array; performing a beamforming processing on the audio signal in a plurality of target directions to obtain a plurality of beam signals; performing a speech recognition on each of the plurality of beam signals to obtain a plurality of speech recognition results corresponding to the plurality of beam signals; and determining a speech recognition result of the audio signal based on the plurality of speech recognition results of the plurality of beam signals.

First claim

What is claimed is:

1. A speech recognition method, performed by an electronic device, the method comprising:
receiving an audio signal obtained by a microphone array;
performing a beamforming processing on the audio signal in a plurality of target directions to obtain a plurality of beam signals by a plurality of beamformers;
performing a speech recognition on each of the plurality of beam signals to obtain a plurality of speech recognition results corresponding to the plurality of beam signals; and
determining a speech recognition result of the audio signal based on the plurality of speech recognition results of the plurality of beam signals,
wherein the performing the speech recognition on each of the plurality of beam signals further comprises:
respectively inputting the plurality of beam signals into corresponding speech recognition models; and
performing the speech recognition on the plurality of beam signals using the speech recognition models in parallel to obtain the plurality of speech recognition results of the plurality of beam signals, and

wherein the plurality of beamformers are divided into one or more groups, each of the one or more groups corresponding to each of the speech recognition models.

2. The method according to claim 1 , wherein the speech recognition result comprises a keyword detection result, and
wherein the determining the speech recognition result of the audio signal further comprises determining a keyword detection result of the audio signal based on a plurality of keyword detection results corresponding to the plurality of beam signals.

3. The method according to claim 2 , wherein the determining the keyword detection result of the audio signal further comprises, based on detecting a keyword in any one of the plurality of beam signals, determining that the keyword is detected in the audio signal.

4. The method according to claim 2 , wherein the keyword detection result comprises a keyword detection probability, and
wherein the determining the keyword detection result further comprises, based on at least one beam signal among the plurality of beam signals being greater than a preset value, determining that the keyword is detected in the audio signal.

5. The method according to claim 2 , wherein the keyword detection result comprises a keyword detection probability, and
wherein the determining the keyword detection result of the audio signal further comprises inputting a plurality of keyword detection probabilities of the plurality of beam signals into a classifier, and determining whether the audio signal includes the keyword based on an output of the classifier.

6. The method according to claim 1 , wherein the determining the speech recognition result of the audio signal further comprises:
obtaining at least one of linguistic scores or acoustic scores of the plurality of speech recognition results; and
determining one of the plurality of speech recognition results having the highest linguistic score or the highest acoustic score as the speech recognition result of the audio signal.

7. The method according to claim 1 , wherein the method further comprises performing a suppression processing on an echo of a second audio signal outputted by a speech recognition device.

8. A speech recognition apparatus, comprising:
at least one memory storing computer program code; and
at least one processor configured to access the at least one memory and operate as instructed by the computer program code, the computer program code comprising:
audio signal receiving code configured to cause the at least one processor to receive an audio signal obtained by a microphone array by a plurality of beamformers;
beamformer code configured to cause the at least one processor to respectively perform a beamforming processing on the audio signal in a plurality of target directions to obtain a plurality of beam signals;
speech recognition code configured to cause the at least one processor to perform a speech recognition on each of the plurality of beam signals to obtain a plurality of speech recognition results corresponding to the plurality of beam signals; and
processing code configured to cause the at least one processor to determine a speech recognition result of the audio signal based on the plurality of speech recognition results of the plurality of beam signals,
wherein the speech recognition code is further configured to cause the at least one processor to:
respectively input the plurality of beam signals into corresponding speech recognition models; and
perform the speech recognition on the plurality of beam signals using the speech recognition models in parallel to obtain the plurality of speech recognition results of the plurality of beam signals, and

wherein the plurality of beamformers are divided into one or more groups, each of the one or more groups corresponding to each of the speech recognition models.

9. The speech recognition apparatus according claim 8 , wherein the processing code is further configured to cause the at least one processor to determine a keyword detection result of the audio signal based on a plurality of keyword detection results corresponding to the plurality of beam signals.

10. The speech recognition apparatus according to claim 9 , wherein the processing code is further configured to cause the at least one processor to, based on detecting a keyword in any one of the plurality of beam signals, determine that the keyword is detected in the audio signal.

11. The speech recognition apparatus according to claim 9 , wherein the processing code is further configured to cause the at least one processor to, based on at least one beam signal among the plurality of beam signals being greater than a preset value, determine that the keyword is detected in the audio signal.

12. The speech recognition apparatus according to claim 8 , wherein the speech recognition code is further configured to cause the at least one processor to:
obtain at least one of linguistic scores or acoustic scores of the plurality of speech recognition results; and
determine one of the plurality of speech recognition results having the highest linguistic score or the highest acoustic score as the speech recognition result of the audio signal.

13. The speech recognition apparatus according to claim 8 , wherein the microphone array comprises at least two annular structures, and
wherein the apparatus further comprises a housing encapsulating the microphone array and the at least one processor.

14. The speech recognition apparatus according to claim 13 , wherein at least three microphones are uniformly disposed on each annular structure.

15. The speech recognition apparatus according to claim 13 , wherein the annular structures are concentric circles.

16. The speech recognition apparatus according to claim 15 , wherein a first microphone and a second microphone on two adjacent annular structures are respectively disposed in the same directions.

17. The speech recognition apparatus according to claim 15 , wherein a first microphone in a first annular structure and a second microphone in a second annular structure are disposed at an angle.

18. The method according to claim 5 , wherein the classifier comprises at least one of a neural network, a support vector machine (SVM), or a decision tree.

19. A non-transitory computer-readable storage medium storing programming code, said programming code configured to cause at least one processor to:
receive an audio signal obtained by a microphone array by a pluralit

Assignees

• Tencent Tech Shenzhen Co Ltd
• Tencent Tech Shenzhen Company Ltd

Inventors

• Gao Yi
• Zheng ji meng
• Yu Meng
• Luo Min

Classifications

• H04R5/027

  Spatial or constructional arrangements of microphones, e.g. in dummy heads · CPC title

• G10L21/0208

  Noise filtering · CPC title

• G10L15/08 Primary

  Speech classification or search · CPC title

• H04R3/005

  for combining the signals of two or more microphones (specially adapted for hearing aids H04R25/407) · CPC title

• H04R3/04

  for correcting frequency response · CPC title