Interactive robot, speech recognition method and computer program product

• US 20060143017A1
• Filed: 12/20/2005
• Published: 06/29/2006
• Est. Priority Date: 12/24/2004
• Status: Active Grant

• Current Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation), Toshiba Digital Solutions Corporation (Toshiba Corporation)
• Original Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation)
• Inventors: Suzuki, Kaoru, Sonoura, Takafumi
• Granted Patent: US 7,680,667 B2
• Time in Patent Office: Days
• US Class Current: 704/275
• CPC Class Codes: G10L 15/26 Speech to text systems G10L15/08 takes precedence

Abstract

An interactive robot capable of speech recognition includes a sound-source-direction estimating unit that estimates a direction of a sound source for target voices which are required to undergo speech recognition; a moving unit that moves the interactive robot in the sound-source direction; a target-voice acquiring unit that acquires the target voices at a position after moving; and a speech recognizing unit that performs speech recognition of the target voices.

Claims

1. An interactive robot capable of speech recognition, comprising:

a sound-source-direction estimating unit that estimates a direction of a sound source for target voices which are required to undergo speech recognition;

a moving unit that moves the interactive robot in the sound-source direction;

a target-voice acquiring unit that acquires the target voices at a position after moving; and

a speech recognizing unit that performs speech recognition of the target voices.

2. The interactive robot according to claim 1, further comprising a voice-producing directing unit by which a sound source of the target voices is directed to produce voices after the robot is moved in the direction of the sound source, wherein the speech recognizing unit performs speech recognition of the target voices produced according to the voice-producing direction.

3. The interactive robot according to claim 1, further comprising:

a signal-to-noise ratio calculating unit that calculates a signal-to-noise ratio of the target voices; and

a signal-to-noise-ratio evaluating unit that compares the calculated signal-to-noise ratio and a predetermined threshold for the signal-to-noise ratio,

wherein the moving unit moves the interactive robot itself in the direction of the sound source when the signal-to-noise ratio is smaller than the threshold for the signal-to-noise ratio.

4. The interactive robot according to claim 3, wherein the target voices are voices produced by an interlocutor communicating with the interactive robot, and

the interactive robot further comprises

an image acquiring unit that acquires images including the interlocutor as the sound source of the target voice;

and

a mouth-movement detecting unit that detects, from the images, mouth movement caused by voices produced by the interlocutor,

wherein the moving unit moves the interactive robot itself in the direction of the sound source when the signal-to-noise ratio is smaller than the threshold for the signal-to-noise ratio, and the mouth movement of the interlocutor is detected.

5. The interactive robot according to claim 3, wherein the target voices are voices produced by an interlocutor communicating with the interactive robot, and

the interactive robot further comprises

an image acquiring unit that acquires images including the interlocutor as the sound source of the target voice;

and

a mouth-movement detecting unit that detects, from the images, mouth movement caused by voices produced by the interlocutor,

wherein the moving unit moves the interactive robot itself in the direction of the sound source when the signal-to-noise ratio is equal to or larger than the threshold for the signal-to-noise ratio, and the mouth movement of the interlocutor is not detected.

6. The interactive robot according to claim 1, further comprising:

a recognition-accuracy evaluating unit that calculates the accuracy of recognition results by the speech recognizing unit,

wherein the moving unit moves the interactive robot itself in the direction of the sound source when the recognition accuracy for results of speech recognition of the target voices is smaller than a predetermined recognition-accuracy threshold.

7. The interactive robot according to claim 6, wherein the target voices are voices produced by an interlocutor communicating with the interactive robot, and

the interactive robot further comprises

an image acquiring unit that acquires images including the interlocutor as the sound source of the target voice;

and

a mouth-movement detecting unit that detects, from the images acquired in the image acquiring unit, mouth movement caused by voices produced by the interlocutor,

wherein the moving unit moves the interactive robot itself in the direction of the sound source when the recognition-accuracy is smaller than the threshold for the recognition accuracy, and the mouth movement of the interlocutor is detected.

8. The interactive robot according to claim 6, wherein the target voices are voices produced by an interlocutor communicating with the interactive robot, and

the interactive robot further comprises

an image acquiring unit that acquires images including the interlocutor as the sound source of the target voice;

and

a mouth-movement detecting unit that detects, from the images, mouth movement caused by voices produced by the interlocutor,

wherein the moving unit moves the interactive robot itself in the direction of the sound source when the recognition-accuracy is equal to or larger than the threshold for the recognition accuracy, and the mouth movement of the interlocutor is not detected.

9. The interactive robot according to claim 6, further comprising a target-voice holding unit that holds a voice pattern of the target voices,

wherein the speech recognizing unit performs speech recognition by pattern matching of the voice pattern of the target voices, which are held in the target-voice holding unit, with the target voices acquired by the target-voice acquiring unit, and

the recognition-accuracy evaluating unit calculates, as the recognition accuracy, an agreement accuracy between the target voices and the voice pattern of the target voices held in the target-voice holding unit.

10. The interactive robot according to claim 9, wherein

the target voices are misrecognition-notification voices signifying that speech recognition by the speech recognizing unit is erroneous,

the target-voice holding unit holds the voice pattern of the misrecognition-notification voices, and

the moving unit moves the interactive robot itself in the direction of the sound source when the misrecognition-notification voices held in the target-voice holding unit are further recognized.

11. The interactive robot according to claim 1, wherein the target voices are voices produced by an interlocutor communicating with the interactive robot,

the interactive robot further comprises

an image acquiring unit that acquires images including the interlocutor as the sound source of the target voice;

and

a mouth-movement detecting unit that detects, from the images acquired in the image acquiring unit, mouth movement caused by voices produced by the interlocutor,

wherein the moving unit moves the interactive robot in the direction of the sound source when the mouth movement is detected and the target voices are not acquired.

12. The interactive robot according to claim 1, wherein the target voices are voices produced by an interlocutor communicating with the interactive robot, and

the interactive robot further comprises

an image acquiring unit that acquires images including the interlocutor as the sound source of the target voice;

and

a mouth-movement detecting unit that detects, from the images, mouth movement of the interlocutor,

wherein the moving unit moves the interactive robot in the direction of the sound source when the mouth movement is not detected and the target voices are acquired.

13. The interactive robot according to claim 1, further comprising

a microphone array that has a plurality of microphones which pick up the target voices,

wherein the direction of the sound source is estimated, based on differential arrival time between the plane waves of the target voices picked up with corresponding voice microphones.

14. The interactive robot according to claim 1, further comprising a distance measuring sensor that measures a distance between the target voices and the interactive robot,

wherein the sound-source-direction estimating unit estimates the direction of the sound source, based on measured results.

15. The interactive robot according to claim 1, further comprising an image forming unit that forms an image of the sound source of the target voices,

wherein the sound-source-direction estimating unit estimates the direction of the sound source, assuming that an image-forming direction is the direction of the sound source.

16. The interactive robot according to claim 1, further comprising:

a signal-strength measurement unit that measures the signal strength of the target voices at a position after the interactive robot is moved by the moving unit; and

an amplification-gain-adjustment unit that, based on the value of the signal strength, adjusts a gain of amplification by which the voice signal of the target voice is amplified, wherein the speech recognizing unit performs speech recognition of the target voices acquired after the gain of amplification is adjusted.

17. A speech recognition method for an interactive robot capable of speech recognition, the method comprising:

estimating a direction of the sound source of target voices which are required to undergo speech recognition;

moving the interactive robot in the direction of the sound source;

acquiring the target voices when the interactive robot is located at a position after moving; and

performing the speech recognition of the target voices.

18. A computer program product having a computer readable medium including programmed instructions for performing speech recognition processing on an interactive robot capable of speech recognition, wherein the instructions, when executed by a computer, cause the computer to perform:

estimating a direction of the sound source of target voices which are required to undergo speech recognition;

moving the interactive robot in the direction of the sound source;

acquiring the target voices when the interactive robot is located-at a position after moving; and

performing the speech recognition of the target voices.