Apparatus, System and Method for Voice Dialogue Activation and/or Conduct

• US 20110145000A1
• Filed: 10/29/2010
• Published: 06/16/2011
• Est. Priority Date: 10/30/2009
• Status: Active Grant

• Current Assignee: Continental Automotive GmbH (Continental AG)
• Original Assignee: Continental Automotive GmbH (Continental AG)
• Inventors: RUEHL, Hans-Wilhelm, KÄMPF, David, HOEPKEN, Harro, KNOBL, Karl-Heinz
• Granted Patent: US 9,020,823 B2
• Time in Patent Office: Days
• US Class Current: 704/275
• CPC Class Codes: G10L 15/06 Creation of reference templates; Training of speech recognition systems, e.g. adaptation to the characteristics of the speaker's voice G10L15/14 takes precedence
• 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 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue
• CPC Class Codes: G10L 17/00 Speaker identification or verification techniques
• CPC Class Codes: G10L 17/26 Recognition of special voice characteristics, e.g. for use in lie detectors; Recognition of animal voices
• CPC Class Codes: G10L 2021/02166 Microphone arrays; Beamforming
• CPC Class Codes: G10L 21/0208 Noise filtering

Abstract

An apparatus, a system and a method for voice dialogue activation and/or conduct. The apparatus for voice dialogue activation and/or conduct has a voice recognition unit, a speaker recognition unit and a decision-maker unit. The decision-maker unit is designed to activate a result action on the basis of results from the voice and speaker recognition units.

Claims

1. An apparatus for at least one of voice dialogue activation and voice dialogue conduct, for use in a vehicle, comprising:

at least one input for a voice signal;

a voice recognition unit configured to establish one or more command words contained in the voice signal;

a speaker recognition unit configured to determine a current speaker using the voice signal and at least one stored speaker profile;

a decision-maker unit comprising;

a voice recognition unit connection coupled to an output of the voice recognition unit configured to perform a result action based on the one or more command words

a speaker recognition unit connection coupled to the speaker recognition unit,

the decision-maker unit configured for the activation of the result action is dependent, at least in the case of at least one command word, on whether the at least one command word has been identified as coming from a speaker associated with a speaker profile.

2. The apparatus as claimed in claim 1, wherein the decision-maker unit is configured to align and correlate results from the speaker recognition unit and from the voice recognition unit with speaker-specific information stored in a speaker profile, wherein performance of at least one command-word-dependent result action is suppressed if a current speaker is not authorized to perform the result actions.

3. The apparatus as claimed in claim 1, wherein the decision-maker unit is configured such that some command words are performed independently of the recognition of a speaker associated with the speaker profile.

4. The apparatus as claimed in claim 1, wherein the apparatus is configured as a combined apparatus for voice dialogue conduct and activation.

5. The apparatus as claimed in claim 1, wherein the voice evaluation unit comprises a word recognition unit configured to recognize words and also a downstream structure evaluation unit configured to recognize command-forming structures.

6. The apparatus as claimed in claim 1, further comprising an echo cancellation unit connected directly or indirectly upstream of at least one of the speaker recognition unit and the voice recognition unit,

wherein the echo cancellation unit has one or more inputs for loudspeaker signals that comprise at least one of mono, stereo, and multichannel loudspeaker signals, the echo cancellation unit configured to compensate for the influence of the loudspeaker signals on the voice signal.

7. The apparatus as claimed in claim 6, wherein the echo cancellation unit comprises a subunit configured to compensate for voice components from other persons, said subunit connected to at least one input for the connection of additional microphones.

8. The apparatus as claimed in claim 1, wherein at least one of the speaker recognition unit and the voice recognition unit has a noise rejection unit connected directly or indirectly upstream.

9. The apparatus as claimed in claim 1, wherein at least one of the speaker recognition unit and the voice recognition unit is configured to synchronize an output from a speaker recognized by the speaker recognition unit to the decision-maker unit with an output of command words recognized by the voice recognition unit.

10. The apparatus as claimed in claim 1, wherein the speaker recognition unit is configured to identify the current speaker by extracting speaker features from the voice signal and comparing the speaker features with stored speaker-dependent features, and comprises a further unit configured for speaker adaptation to continually ascertain refined speaker-dependent features and store the refined speaker-dependent features in the stored speaker profiles.

11. The apparatus as claimed in claim 1, wherein a driver state sensing unit for sensing a state of the driver using the voice signal is arranged in parallel with the speaker recognition unit and the voice recognition unit.

12. The apparatus as claimed in claim 1, wherein the voice recognition unit comprises an additional unit configured to capture time-related alterations in the speaker features of a speaker as an attribute and to store them in a stored speaker profile associated with the speaker.

13. The apparatus as claimed in claim 1, further comprising at least one memory apparatus configured for storing at least one of user profiles and speaker profiles.

14. The apparatus as claimed in claim 13, wherein the at least one memory apparatus has one of an input and output for one of exporting and importing the one of the stored speaker profiles and the user profiles.

15. The apparatus as claimed in claim 15, wherein the apparatus is activated to evaluate the voice signals even during the performance of a result action, such that recognition of a command from an authorized speaker prompts at least partial interruption of the performance of a result action triggered by a prior command.

16. A system for voice dialogue activation and/or voice dialogue conduct comprising:

at least one input for a voice signal;

a voice recognition unit configured to establish one or more command words contained in the voice signal;

a speaker recognition unit configured to determine a current speaker using the voice signal and at least one stored speaker profile;

a decision-maker unit comprising;

a voice recognition unit connection coupled to an output of the voice recognition unit configured to perform a result action based on the one or more command words

a speaker recognition unit connection coupled to the speaker recognition unit,

the decision-maker unit configured for the activation of the result action is dependent, at least in the case of at least one command word, on whether the at least one command word has been identified as coming from a speaker associated with a speaker profile;

at least one microphone coupled to the voice recognition unit; and

at least one loudspeaker coupled to the voice recognition unit.

17. The system as claimed in claim 16, further comprising at least one of a plurality of microphones and at least one microphone array arranged such that areas of optimum reception provided by directional characteristics of the microphones, for at least some of the microphones overlap in the presumed area of authorized speakers.

18. The system as claimed in claim 17, wherein the microphones are configured to automatically orient to a position of the speaker sensed by the microphones.

19. A method for voice dialogue activation and/or conduct comprising:

picking up a voice signal;

recognizing at least one of a command word and a command word structure from the voice signal;

recognizing a speaker using the voice signal and at least one stored speaker profile; and

performing a result action based on a recognized command word and a recognized speaker.

20. The method as claimed in claim 19, wherein the recognizing the at least one of the command word or the command word structure contains further comprises:

recognizing words contained in the voice signal; and

recognizing command structures formed by the words.

21. The method as claimed in claim 19, wherein at least one of the recognition of the command word and the recognition of a speaker is preceded by performance of echo cancellation such that overlays from a loudspeaker signal produced by reflections in a passenger compartment are removed from the voice signal by calculating the overlays produced by the loudspeaker signal and subtracting them from the voice signal.

22. The method as claimed in claim 21, wherein voice signal components of the voice signal by further persons are determined and at least partially removed from the voice signal.

23. The method as claimed in claim 22, wherein the voice signal is a multichannel voice signal.

24. The method as claimed in claim 23, further comprising chronological correlation of differently time-shifted signals from different channels of the multichannel voice signal to extract and separate those components of the voice signal which come from one of the locations of authorized speakers.

25. The method as claimed in claim 19, wherein a dependency of performance of a result action on a recognized command word and a recognized speaker involves performance of a result action being suppressed if the associated speaker one of has not been recognized and is not authorized to instruct the result action.

26. The method as claimed in claim 19, wherein the recognition of one of the command word or of the speaker is preceded by performance of noise rejection.

27. The method as claimed in claim 19, wherein recognizing an authorized speaker involves speaker features being extracted from the voice signal and being aligned with individual speaker features stored in a speaker profile.

28. The method as claimed in claim 27, wherein speaker adaptation is performed which continuously refines and complements the individual speaker features stored in the speaker profile.

29. The method as claimed in claim 28, wherein time-related alterations in the speaker features are captured as an attribute and stored in the speaker profile.

30. The method as claimed in claim 29, wherein the recognizing command words comprises extracting voice features from the voice signal (8;

37 aligned with individual voice features stored in the speaker profile.

31. The method as claimed in claim 30, wherein speaker adaptation is performed which continuously refines and complements the individual voice features stored in the speaker profile.

32. The method as claimed in claim 31, wherein time-related alterations in the voice features are captured as an attribute and stored in the speaker profile.

33. The method as claimed in claim 32, wherein the voice signal is used to sense a driver state.

34. The method as claimed in claim 19, wherein subsequent performance of the result action is interrupted by input of a further voice signal, containing a further command word from an authorized speaker.

35. The method as claimed in claim 19, wherein the voice signal is used to ascertain locations of authorized speakers comprising producing a control signal for orienting at least one microphone to locations of the authorized speakers independent of a command word contained in the voice signal.

36. The method as claimed in claim 35, wherein the performance of the result action involves output of a voice dialogue signal.

37. The method as claimed in claim 36, wherein the performance of the result action comprises signal output of a control signal to controls a function of an apparatus integrated in a vehicle.