Signal injection coupling into the human vocal tract for robust audible and inaudible voice recognition

• US 6,487,531 B1
• Filed: 07/06/1999
• Issued: 11/26/2002
• Est. Priority Date: 07/06/1999
• Status: Expired due to Term

• Current Assignee: Carol A. Tosaya, John W. Sliwa Jr
• Original Assignee: Carol A. Tosaya, John W. Sliwa Jr
• Inventors: Tosaya, Carol A., Sliwa, Jr., John W.
• Primary Examiner(s): Banks-Harold, Marsha D.
• Assistant Examiner(s): ABEBE, DANIEL DEMELASH
• Application Number:
  US09/348,600
• Time in Patent Office: 1,239 Days
• Field of Search: 704/246, 704/223, 704/212, 704/251, 704/207, 704/264, 704/208
• US Class Current: 704/246
• 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 19/08 Determination or coding of the excitation function; Determination or coding of the long-term prediction parameters

Abstract

A means and method are provided for enhancing or replacing the natural excitation of the human vocal tract by artificial excitation means, wherein the artificially created acoustics present additional spectral, temporal, or phase data useful for (1) enhancing the machine recognition robustness of audible speech or (2) enabling more robust machine-recognition of relatively inaudible mouthed or whispered speech. The artificial excitation (a) may be arranged to be audible or inaudible, (b) may be designed to be non-interfering with another user'"'"'s similar means, (c) may be used in one or both of a vocal content-enhancement mode or a complimentary vocal tract-probing mode, and/or (d) may be used for the recognition of audible or inaudible continuous speech or isolated spoken commands.

Claims

1. A speech recognition system for processing sounds emanating from a living body'"'"'s vocal tract, said sounds including sounds excited by at least one artificial exciter coupled, either directly or indirectly, into said vocal tract to introduce artificial excitations, said at least one artificial excitation modified or modulated by said vocal tract and emanating therefrom, said speech recognition system including:

means for representation, modeling or classification or both, and searching of artificially excited speech signals or signal components;

means for representation, modeling or classification or both, and searching of naturally excited speech signals or signal components;

at least one said searching means having access to at least one of an acoustic model, lexical model or language model;

at least one training means; and

means for directing at least a first modified or modulated artificially excited speech signal to a first speech representation means which samples at least said first signal to produce a first sequence of speech representation vectors, representative at least in part, of said artificially excited signal, wherein both the artificially excited signal and the naturally excited signal are represented by a single set of representation vectors.

2. The speech recognition system of claim 1 wherein said sounds are one of continuous speech, command-style speech, or an utterance.

3. The speech recognition system of claim 1 further including means for modeling or classifying said first sequence of vectors.

4. The speech recognition system of claim 3 further including means for subjecting said modeled or classified vectors to a search in a search module, said search module having access to at least one of an acoustic model, a lexical model, or a language model.

5. The speech recognition system of claim 4 wherein two search modules operate, one arranged to process naturally excited signals and the other to process artificially excited signals, said system utilizing the results of both modules to decide what speech took place or what words were articulated.

6. The speech recognition system of claim 1 further including means for directing at least a naturally excited second modified or modulated signal to a speech representation means which samples said naturally excited signal to produce a second sequence of speech representation vectors, representative at least in part of said natural speech signal.

7. The speech recognition system of claim 6 further including second means for modeling or classifying said second sequence of vectors representative, at least in part, of said naturally excited speech signal.

8. The speech recognition system of claim 7 further including second means for subjecting said modeled or classified natural speech vectors to a search in a second search module said search module having access to at least one of an acoustic model, a lexical model or a language model.

9. The speech recognition system of claim 1 wherein training means are provided for both naturally excited signals and artificially excited signals, said means being one of independent or the same means, said signals being one of separate or combined.

10. The speech recognition system of claim 1 wherein artificial excitations are adapted to an individual user.

11. The speech recognition system of claim 10 wherein said adapted excitations are portable across at least one of multiple recognition systems, computers, networks, and speech-conversant devices.

12. The speech recognition system of claim 1 further including a separator, deconvolution, or subtraction means to discern naturally excited sounds or sound components from artificially excited sounds or sound components.

13. The speech recognition system of claim 1 wherein said artificially excited sounds permit inaudible speaking or command-giving to a computer, computer-coupled device, or computer-containing device.

14. The speech recognition system of claim 1 adapted for processing sounds that are both naturally excited and artificially excited, said sounds, or signal representations thereof, being substantially processed as one of separate or separated signals or signal-components or as a combined signal.

15. The speech recognition system of claim 14 wherein said artificially excited sounds permit improved recognition-accuracy or improved recognition-speed of natural speech, sounds or utterances.

16. The speech recognition system of claim 14 wherein said artificially excited and naturally excited speech sounds emanating from said tract temporally overlap at least part of the time.

17. The speech recognition system of claim 14 wherein said artificially excited and naturally excited speech sounds emanating from said tract are not identical in spectral content at least part of the time.

18. The speech recognition system of claim 14 wherein said artificially excited signal, before or after tract modification or modulation, includes at least one of the following aspects:

(a) said artificially excited signal contains a harmonic or subharmonic of a natural formant, (b) said artificially excited signal contains phase information which is utilized in the recognizer, (c) said artificially excited signal is broadband in nature, (d) said artificially excited signal is selected or set as a function of any natural signal parameter, (e) said artificially excited signal contains tones or frequency components which interact with each other as a function of a vocal tract parameter, (f) said artificially excited signal contains at least one tone or frequency component which is modulated or modified by any portion of the vocal tract anatomy, (g) said artificially excited signal is generally inaudible to the unaided ear of a separate listener, and (h) said artificially excited signal is swept in frequency.

19. The speech recognition system of claim 1 wherein said vocal tract includes at least one element selected from the group consisting of vocal chords, larynx, laryngeal valve, the glottal opening, the glottis, the arytenoids, the pharynx, the esophagus, the tongue, the pharyngeal walls, the velum, the hard palate, the alveolar ridge, the lips, teeth, gums, cheeks or any nasal cavity, at least said one element modifying or modulating said artificial excitation as the speaker articulates speech either audibly or inaudibly.

20. The speech recognition system of claim 1 further including a training data means capable of supporting training using at least the artificially excited speech signals.

21. The speech recognition system of claim 1 wherein said at least one artificial excitation is chosen based on an optimized correlation between it and known words or utterances made available during training.

22. A method of minimizing degradation in the accuracy or speed of speech-recognition of a first speaker'"'"'s speech or utterance caused by at least one second interfering background speaker, voice, or sound, said method comprising:

coupling artificial acoustic excitation, directly or indirectly, into the vocal tract of the first speaker;

allowing said first speaker to audibly speak in the potential acoustic presence of said at least one second background speaker or sound, thereby modifying or modulating said first speaker'"'"'s artificial acoustic excitation as well as said first speaker'"'"'s natural excitation; and

processing at least a portion of said first speaker'"'"'s artificially-produced acoustic output by a speech recognition means, said speech recognition means comprising;

means for representation, modeling or classification, and searching of artificially excited speech signals or signal components;

means for representation, modeling or classification, and searching of naturally excited speech signals or signal components;

at least one of said searching means having access to at least one of an acoustic model, lexical model or language model; and

at least one training means;

wherein said first speaker'"'"'s output is known to be that of said first speaker due to its identifiable artificial acoustic content,

or wherein said second speaker'"'"'s or sound'"'"'s interfering output is ignored or rejected because it does not contain first speakers identifying artificial excitations.

23. The method of claim 22 wherein at least two said equipped speakers are one of (a) speaking as part of a conversing group of at least two or (b) speaking to each other locally or from remote locations.

24. The method of claim 22 wherein speech recognition means process at least portions of both naturally-excited and artificially-excited output of said speaker.

25. The method of claim 24 wherein temporally and/or spectrally unique artificial excitations are provided to two or more thus-equipped speakers such that all such equipped speakers may speak and be recognized without recognition interference with each other, said unique excitations as sociable with particular speakers.

26. The method of claim 25 wherein a thus-equipped speaker'"'"'s recognition system is arranged to ignore or reject inputs containing modifications of, modulations of, or elements of a potentially interfering speaker'"'"'s different artificial excitation and audible speech as sociable with said interfering speaker.

27. The method of claim 25 wherein a computer provides or assigns said unique artificial excitations.

28. The method of claim 27 wherein information regarding at least one unique artificial excitation, or assignment thereof, is delivered by one of a computer network, telecommunications network, wireless signal, or is inputted manually or via speech-input.

29. The method of claim 22 further comprising:

choosing said at least one artificial excitation based on an optimized correlation between it and known words or utterances made available during training.

30. A method of providing a speech-recognition based security function for user identification or validation comprising:

(a) coupling, directly or indirectly, an artificial acoustic exciter into a user'"'"'s vocal tract;

(b) having the user speak, articulate or mouth an utterance wherein said utterance, at least in part, comprises a portion of the artificial excitation as-modified or modulated by said user'"'"'s vocal tract;

(c) applying speech recognition processing means to identify or validate said user, said means processing at least a portion of said artificially excited speech, utterance or signal-representation thereof; and

(d) storing information relating to at least one characteristic of said user'"'"'s vocal tract, or of its function, being used in said user identification or validation process, wherein said speech-recognition processing includes processing said modified acoustic excitation through representation, modeling or classification or both, and searching to produce identified words.

31. The method of claim 30 wherein said user speaks or utters at least one designated entry-utterance for the purpose of said identification or validation, said audible or inaudible entry-utterance comprising at least one of:

(a) including at least a portion of said user'"'"'s name or alias;

(b) including a welcoming greeting;

(c) being revealed to said user only at the time of attempted entry; and

(d) being revealed to said user after its random selection.

(4) Improves security for speech-based user-identification or user-validation.

32. The method of claim 30 further comprising:

choosing said at least one artificial excitation based on an optimized correlation between it and known words or utterances made available during training.