Speech recognition method and speech recognition device

• US 9,818,403 B2
• Filed: 10/27/2015
• Issued: 11/14/2017
• Est. Priority Date: 08/29/2013
• Status: Active Grant

• Current Assignee: Panasonic Intellectual Property Corporation of America (Panasonic Holdings Corporation)
• Original Assignee: Panasonic Intellectual Property Corporation of America (Panasonic Holdings Corporation)
• Inventors: Nishikawa, Tsuyoki
• Primary Examiner(s): Vo, Huyen
• Application Number:
  US14/923,912
• Publication Number:
  US 20160049150A1
• Time in Patent Office: 749 Days
• Field of Search: 704230, 704231, 704233, 704251, 704250, 704255, 704257, 704270, 704275
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue
• CPC Class Codes: G10L 2015/223 Execution procedure of a spoken command
• CPC Class Codes: G10L 2021/02166 Microphone arrays; Beamforming

Abstract

A speech recognition method is provided that recognizes speech for causing equipment to operate. The method includes acquiring a speech signal from a microphone disposed in a designated space. The method also includes detecting a spatial sound pressure distribution indicating a distribution of sound pressure in the space, on the basis of the acquired speech signal, and detecting a point sound source in the space on the basis of the detected spatial sound pressure distribution. The method further includes judging to conduct a speech recognition process on the acquired speech signal when the point sound source is detected.

Claims

1. A speech recognition method that recognizes speech for causing equipment to operate, comprising:

acquiring speech signal from a microphone disposed in a designated space;

detecting a point sound source in the designated space on the basis of the acquired speech signal;

judging to conduct a speech recognition process on the acquired speech signal when the point sound source is detected by the detecting of the point sound source; and

detecting a spatial sound pressure distribution indicating a distribution of sound pressure in the designated space on the basis of the acquired speech signal,

wherein in the detecting of the point sound source, the point sound source is detected on the basis of the spatial sound pressure distribution,

wherein the spatial sound pressure distribution is a sound pressure distribution at designated angles with reference to a position of the microphone, and

wherein the detecting of the point sound source includes

specifying an angular range of one or more sound sources for which sound pressure in the speech signal is higher than a designated threshold, on the basis of the spatial sound pressure distribution, and

detecting, from among the specified one or more sound sources, a sound source having a width of angular range that is less than or equal to a designated width as a point sound source.

2. The speech recognition method according to claim 1,

wherein the judging, on the basis of the spatial sound pressure distribution, judges to conduct a speech recognition process on a sound source detected in a region within a designated angle from the microphone, and judges not to conduct a speech recognition process on a sound source detected in a region outside a designated angle from the microphone.

3. The speech recognition method according to claim 1, further comprising:

specifying a position in the designated space of the detected point sound source when the point sound source is detected in the detecting of a point sound source,

wherein when a distance between the position of the point sound source and the equipment is shorter than a designated distance, the judging judges to conduct the speech recognition process, whereas when the distance between the position of the point sound source and the equipment is equal to or greater than the designated distance, the judging judges not to conduct the speech recognition process.

4. The speech recognition method according to claim 3, further comprising:

controlling a directionality of the microphone to point towards the specified position of the point sound source.

5. The speech recognition method according to claim 1, further comprising:

generating output audio information to be output from a speaker;

outputting the generated output audio information as output audio from the speaker; and

removing the generated output audio information from the acquired speech signal as an echo component.

6. The speech recognition method according to claim 1, further comprising:

specifying a position in the designated space of the detected point sound source when the point sound source is detected by the detecting of the point sound source; and

specifying a direct-arrival point sound source from which emitted sound arrives at the microphone directly for the point sound source whose position is specified.

7. The speech recognition method according to claim 6,

wherein the judging judges to conduct the speech recognition process when the direct-arrival point sound source is specified in the specifying of the direct-arrival point sound source, and judges not to conduct the speech recognition process when the direct-arrival point sound source is not specified in the specifying of the direct-arrival point sound source.

8. The speech recognition method according to claim 3, further comprising:

specifying a direct-arrival point sound source from which emitted sound arrives at the microphone directly for the point sound source whose position is specified; and

controlling a directionality of the microphone to point towards the specified position of the point sound source when the direct-arrival point sound source is specified.

9. The speech recognition method according to claim 6, further comprising:

identifying whether or not the directly arriving sound is speech spoken by a person when the direct-arrival point sound source is specified,

wherein the judging judges to conduct the speech recognition process when the directly arriving sound is identified as being speech spoken by a person, and judges not to conduct the speech recognition process when the directly arriving sound is identified as not being speech spoken by a person.

10. The speech recognition method according to claim 1, further comprising:

when the speech recognition process is judged to be conducted in the judging, reading out at least one word related to operating the equipment from a storage storing in advance a plurality of words used in the speech recognition process, and displaying the at least one read-out word on a display.

11. The speech recognition method according to claim 1, further comprising:

increasing a brightness of a display when the judging judges to conduct the speech recognition process over a brightness of a display unit when the judging judges not to conduct the speech recognition process.

12. The speech recognition method according to claim 1, further comprising:

displaying on a display an animation image according to a judgment of whether or not to conduct the speech recognition process in the judging.

13. A speech recognition device that recognizes speech for causing equipment to operate, comprising:

one or more memories; and

circuitry that, in operation

acquires speech signal from a microphone disposed in a designated space;

detects a point sound source in the designated space on the basis of the acquired speech signal;

judges to conduct a speech recognition process on the acquired speech signal when the point sound source is detected; and

detects a spatial sound pressure distribution indicating a distribution of sound pressure in the designated space on the basis of the acquired speech signal,

wherein the point sound source is detected on the basis of the spatial sound pressure distribution,

wherein the spatial sound pressure distribution is a sound pressure distribution at designated angles with reference to a position of the microphone, and

wherein detecting of the point sound source includes

specifying an angular range of one or more sound sources for which sound pressure in the speech signal is higher than a designated threshold, on the basis of the spatial sound pressure distribution, and

detecting, from among the specified one or more sound sources, a sound source having a width of angular range that is less than or equal to a designated width as a point sound source.