Speech recognition using microphone antenna array

• US 6,937,980 B2
• Filed: 10/02/2001
• Issued: 08/30/2005
• Est. Priority Date: 10/02/2001
• Status: Active Grant

• Current Assignee: Optis Cellular Technology LLC (Brevet Capital)
• Original Assignee: Telefonaktiebolaget LM Ericsson
• Inventors: Makovicka, Thomas, Krasny, Leonid, Khayrallah, Ali
• Primary Examiner(s): Ometz, David L.
• Assistant Examiner(s): Opsasnick, Michael N.
• Application Number:
  US09/970,369
• Publication Number:
  US 20030069727A1
• Time in Patent Office: 1,428 Days
• Field of Search: 704/231, 704/233, 704/270, 704/275
• US Class Current: 704/231
• CPC Class Codes: G10L 15/20 Speech recognition techniques specially adapted for robustness in adverse environments, e.g. in noise, of stress induced speech G10L21/02 takes precedence
• CPC Class Codes: G10L 2021/02166 Microphone arrays; Beamforming
• CPC Class Codes: G10L 2021/02168 the estimation exclusively taking place during speech pauses
• CPC Class Codes: G10L 21/0208 Noise filtering

Abstract

A system and method of audio processing provides enhanced speech recognition. Audio input is received at a plurality of microphones. The multi-channel audio signal from the microphones may be processed by a beamforming network to generate a single-channel enhanced audio signal, on which voice activity is detected. Audio signals from the microphones are additionally processed by an adaptable noise cancellation filter having variable filter coefficients to generate a noise-suppressed audio signal. The variable filter coefficients are updated during periods of voice inactivity. A speech recognition engine may apply a speech recognition algorithm to the noise-suppressed audio signal and generate an appropriate output. The operation of the speech recognition engine and the adaptable noise cancellation filter may advantageously be controlled based on voice activity detected in the single-channel enhanced audio signal from the beamforming network.

Claims

1. A method of recognizing speech, comprising:

receiving speech input via a plurality of microphones;

processing a corresponding plurality of audio signals from said microphones with a beamforming network to generate a first improved audio signal;

detecting voice activity on said first improved audio signal;

processing said plurality of audio signals from said microphones with an adaptive noise cancellation filter having variable filter coefficients to generate a second improved audio signal;

repeatedly updating said variable filter coefficients during periods of voice inactivity as detected on said first improved audio signal to minimize non-speech audio components in said second improved audio signal; and

performing speech recognition on said second improved audio signal.

2. The method of claim 1, wherein performing speech recognition on said second improved audio signal occurs in response to detecting voice activity on said first improved audio signal.

3. The method of claim 1 wherein said detecting voice activity on said first improved audio signal comprises detecting voice activity on an improved audio signal neither generated by said adaptive noise cancellation filter nor derived from any signal output thereby.

4. A method of speech recognition, comprising:

receiving audio input at a microphone array comprising a plurality of microphones and generating a multi-channel audio signal;

processing said multi-channel audio signal into a first single channel audio signal by a beamforming network;

detecting voice activity in said first single channel audio signal;

processing said multi-channel audio signal into a second single channel audio signal by an adaptable audio filter;

selectively updating parameters associated with said adaptable audio filter;

timing said updating of parameters based on said voice activity detection; and

performing speech recognition on said second single channel audio signal.

5. The method of claim 4, further comprising timing said speech recognition based on said voice activity detection.

6. A system for speech recognition, comprising:

a microphone array comprising a plurality of microphones for receiving audio input, said microphone array generating a multi-channel audio signal;

a beamforming network receiving said multi-channel audio signal and generating a first single channel audio signal;

a voice activity detection module receiving said first single channel audio signal and generating a binary signal indicative of the presence or absence of a voice component in said first single channel audio signal;

an adaptive noise cancellation filter having variable filter coefficients receiving said multi-channel audio signal and said binary signal, and generating a second single channel audio signal, said filter coefficients being updated in response to said binary signal;

a speech recognition engine receiving said second single channel audio signal, operative to interpret speech in said second single channel audio signal in response to said binary signal.

7. The system of claim 6, wherein said filter coefficients are updated in response to said binary signal indicating the absence of a voice component in said first single channel audio signal.

8. The system of claim 6, wherein said speech recognition engine additionally receives said binary signal, and interprets speech in said second single channel audio signal in response to said binary signal indicating the presence of a voice component in said first single channel audio signal.

9. The system of claim 6, wherein said beamforming network, said voice activity detection module, said adaptive noise cancellation filter, and said speech recognition engine are included in a wireless communications mobile terminal.

10. The system of claim 6, wherein one or more of said beamforming network, said voice activity detection module, said adaptive noise cancellation filter, and said speech recognition engine are included in a hands-free adapter capable of connecting to a wireless communications mobile terminal.

11. The system of claim 6, wherein one or more of said beamforming network, said voice activity detection module, said adaptive noise cancellation filter, and said speech recognition engine are implemented in software.