User selectable multiple threshold criteria for voice recognition

• US 5,960,393 A
• Filed: 06/12/1997
• Issued: 09/28/1999
• Est. Priority Date: 07/31/1995
• Status: Expired due to Term

• Current Assignee: Alcatel-Lucent USA, Inc. (Nokia Corporation)
• Original Assignee: Lucent Technologies, Inc. (Nokia Corporation)
• Inventors: Keen, Donald Marion, Keen, Ellen Anne, Deldar, Mitra P., Cohrs, Paul Wesley
• Primary Examiner(s): Hudspeth, David R.
• Assistant Examiner(s): Storm, Donald L.
• Application Number:
  US08/873,613
• Time in Patent Office: 838 Days
• Field of Search: 704/200, 704/236-240, 704/231, 704/246, 704/270, 704/275, 704/251, 704/233, 704/247, 704/250, 704/252
• US Class Current: 704/240
• CPC Class Codes: G10L 15/08 Speech classification or search
• CPC Class Codes: G10L 15/10 using distance or distortion measures between unknown speech and reference templates
• 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 15/28 Constructional details of speech recognition systems
• CPC Class Codes: G10L 2015/223 Execution procedure of a spoken command
• CPC Class Codes: G10L 2015/227 of the speaker; Human-factor methodology
• CPC Class Codes: G10L 2025/783 based on threshold decision
• CPC Class Codes: H04M 1/64 Automatic arrangements for answering calls; Automatic arrangements for recording messages for absent subscribers; Arrangements for recording conversations centralised dictation systems H04M11/10
• CPC Class Codes: H04M 2250/74 with voice recognition means voice activated dialling H04M1/271; voice recognition algorithms G10L15/00

Abstract

A method and apparatus for speech recognition in which a single criterion or set of criteria is selected manually by the user, from plural classes of recognition criteria. The stored classes of recognition criteria include a default class optimized for an average user in normal conditions, at least one class having a probability of recognition greater than said default class, and at least one class having a probability of recognition less than said default class. Accordingly, the user may select that class of criteria which provides the best results for him or her, as measured by greater accuracy (fewer false positive detections) or fewer instances of non-rejection. An utterance is compared to one or more models of speech to determine a similarity metric for each such comparison. The model of speech which most closely matches the utterance is determined based on the one or more similarity metrics. The similarity metric corresponding to the most closely matching model of speech is analyzed to determine whether the similarity metric satisfies the criteria of the user-selected class. The present application has application to many problems in speech recognition including isolated word recognition and command spotting. Illustrative embodiments of the invention in the context of telecommunications instruments are provided.

Claims

1. A speech recognizer apparatus for recognizing a phrase including at least one word, based upon an utterance, the apparatus comprising:

a plurality of stored recognition criteria sets, each stored recognition criteria set having at least one class of recognition criteria, each stored class being associated with a corresponding probability of recognition for a given utterance;

a selection module responsive to a user, wherein said user selects a selected recognition criteria set from said plurality of stored recognition criteria sets;

a comparator module, responsive to the selection module, for determining whether a similarity metric corresponding to a most closely matching model of speech satisfies the selected recognition criteria set; and

a recognizer module for recognizing the utterance as the phrase corresponding to said most closely matching model of speech when the selected recognition criteria set is satisfied.

2. The apparatus of claim 1, wherein a model of speech comprises one or more predetermined words.

3. The apparatus of claim 2, wherein a predetermined word comprises a command word for a utilization device.

4. The apparatus of claim 3, wherein said selection is further responsive to said plurality of users wherein each of said users selects a selected recognition criteria set from said plurality of stored recognition criteria sets on a per command word basis.

5. The apparatus of claim 4, wherein said selection is further response to said plurality of users wherein each of said user selects a selected recognition criteria set from said plurality of stored recognition criteria sets on a per command word family basis.

6. The apparatus of claim 2, further including a utilization device for incorporating said speech recognizer apparatus.

7. The apparatus of claim 6, wherein the utilization device is a telephone.

8. The apparatus of claim 7, wherein the utilization device is a wireless telephone.

9. The apparatus of claim 6, wherein the utilization device is an answering machine.

10. The apparatus of claim 2, wherein each recognition criteria set comprises a set of thresholds and wherein the comparator module is operative to compare the similarity metric corresponding to said most closely matching model of speech to a set of thresholds for said selected recognition criteria set.

11. The apparatus of claim 10, wherein said stored recognition criteria sets includes a default class optimized for an average user in normal conditions, at least one class having a probability of recognition greater than said default class, and at least one class having a probability of recognition less than said default class.

12. The apparatus of claim 1, wherein said selection module is responsive to a plurality of users wherein each of said users selects a selected recognition criteria set from said plurality of stored recognition criteria sets on a per user basis.

13. The apparatus of claim 12, wherein said selection module is further responsive to said plurality of users wherein each of said users selects a selected recognition criteria set from said plurality of stored recognition criteria sets on a per utterance basis.

14. A telecommunications instrument, comprising:

a microphone; and

a speech recognizer for recognizing a phrase including at least one word, based upon a utterance, the speech recognizer including

a plurality of stored recognition criteria sets, each stored criteria set having at least one class of recognition criteria, each stored class being associated with a corresponding probability of recognition for a given utterance;

a selection module response to a user, wherein said user selects a selected recognition criteria set from said plurality of stored recognition criteria sets;

a comparator module, responsive to the selecting module, for determining whether a similarity metric corresponding to a most closely matching model of speech satisfies the selected recognition criteria set; and

a device control circuit responsive to recognition of an utterance by said speech recognizer.

15. The telecommunications instrument of claim 14, wherein said device control circuit is a telephone circuit for providing telephone operation in response to recognition of an utterance.

16. The telecommunications instrument of claim 15, further comprising:

a radio transceiver; and

an audio processor for interfacing the microphone and the telephone circuit to the transceiver, the audio processor being responsive to control signals provided by the telephone circuit.

17. The telecommunications instrument of claim 14, wherein said device control circuit is answering machine circuit for audibly reproducing stored messages in response to recognition of an utterance.

18. The telecommunications instrument of claim 17, wherein the microphone is a hands-free microphone.

19. The telecommunications instrument of claim 17, wherein the microphone is a handset microphone of a cordless telephone.

20. The telecommunications instrument of claim 14, further comprising a radio transceiver coupled to said device control circuit to provide wireless telephone communication.

21. The telecommunications instrument of claim 14, wherein the microphone is a hands-free microphone.

22. A method of recognizing a phrase including at least one word, based upon an utterance, the method comprising the steps of

comparing the utterance to one or more speech models to determine a similarity metric for each such comparison;

determining, in a first determining step, which model of speech most closely matches the utterance based on the one or more similarity metrics obtained during said comparing step;

selecting, by a user, at least one recognition criteria set from a plurality of stored recognition criteria sets each stored criteria set having at least one class of recognition criteria, each stored class being associated with a corresponding probability of recognition for a given utterance;

determining, in a second determining step, whether the similarity metric corresponding to the most closely matching model of speech satisfies the selected recognition criteria set; and

recognizing the utterance as the phrase corresponding to said most closely matching model of speech when the selected recognition criteria set is satisfied.