Updating markov models based on speech input and additional information for automated telephone directory assistance

• US 5,644,680 A
• Filed: 05/25/1995
• Issued: 07/01/1997
• Est. Priority Date: 04/14/1994
• Status: Expired due to Term

• Current Assignee: Volt Delta Resources LLC (NewNet Communication Technologies LLC)
• Original Assignee: Northern Telecom Limited (Nortel Networks Corporation)
• Inventors: Sharp, R. Douglas, Lennig, Matthew, Wasmeier, Hans A., Hodgson, Lauren C., Bielby, Gregory J., Gupta, Vishwa N.
• Primary Examiner(s): MacDonald, Allen R.
• Assistant Examiner(s): MATTSON, ROBERT
• Application Number:
  US08/452,191
• Time in Patent Office: 768 Days
• Field of Search: 381/43, 371/67, 395/2.45, 395/2.46, 395/2.48, 395/2.49, 395/2.53, 395/2.65
• US Class Current: 704/240
• CPC Class Codes: G10L 15/063 Training
• CPC Class Codes: G10L 15/142 Hidden Markov Models [HMMs]
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue
• CPC Class Codes: G10L 25/24 the extracted parameters being the cepstrum
• CPC Class Codes: H04M 2201/40 using speech recognition
• CPC Class Codes: H04M 2242/22 Automatic class or number identification arrangements
• CPC Class Codes: H04M 3/42093 Notifying the calling party of information on the called or connected party
• CPC Class Codes: H04M 3/42102 Making use of the called party identifier
• CPC Class Codes: H04M 3/4931 Directory assistance systems
• CPC Class Codes: H04M 3/51 Centralised call answering arrangements requiring operator intervention , e.g. call or contact centers for telemarketing
• CPC Class Codes: H04Q 3/72 Finding out and indicating number of calling subscriber

Abstract

In methods and apparatus for at least partially automating a telephone directory assistance function, directory assistance callers are prompted to speak locality or called entity names associated with desired directory listings. A speech recognition algorithm is applied to speech signals received in response to prompting to determine spoken locality or called entity names. Desired telephone numbers are released to callers, and released telephone numbers are used to confirm or correct at least some of the recognized locality or called entity names. Speech signal representations labelled with the confirmed or corrected names are used as labelled speech tokens to refine prior training of the speech recognition algorithm. The training refinement automatically adjusts for deficiencies in prior training of the speech recognition algorithm and to long term changes in the speech patterns of directory assistance callers served by a particular directory assistance installation. The methods can be generalized to other speech recognition applications.

Claims

1. A method for performing desired actions in response to speech signals, comprising:

storing representations of speech signals;

calculating, according to a speech recognition algorithm responsive to the representations of speech signals, measures of probability that the speech signals correspond to each of a plurality of actions in an action vocabulary;

selecting actions in response to the calculated measures of probability and automatically performing said actions from the action vocabulary;

acquiring further data indicative of desired actions;

calculating, according to a speech recognition algorithm responsive to both the representations of the speech signals and the further data, further measures of probability that the speech signals correspond to each of a plurality of actions;

labelling the stored representations of speech signals in response to the further calculated measures of probability; and

calculating speech recognition algorithm model parameters in response to the labelled stored representations of speech signals.

2. A method as defined in claim 1, wherein at least some of the selected actions comprise providing selected items of desired information.

3. A method as defined in claim 1, wherein:

the step of automatically performing selected actions comprises prompting speakers to provide further speech signals indicative of desired actions; and

the step of acquiring further data comprises calculating, according to a speech recognition algorithm responsive to the further speech signals, measures of probability that the speech signals correspond to each of a plurality of actions in an action vocabulary.

4. A method as defined in claim 3, wherein the step of prompting speakers to provide further speech signals comprises prompting speakers for disconfirmation of desired actions selected in response to previously analyzed speech signals.

5. A method as defined in claim 4, wherein the step of prompting speakers for disconfirmation is performed selectively dependent on the particular actions selected in response to previously analyzed speech signals.

6. A method as defined in claim 1, wherein the step of acquiring further data indicative of desired actions comprises monitoring for operator-initiated disconfirmations of actions selected in response to previously analyzed speech signals.

7. A method as defined in claim 6, wherein the step of monitoring for operator-initiated disconfirmations comprises monitoring for manual over-rides of actions selected in response to previously analyzed speech signals.

8. A method as defined in claim 6, wherein the step of monitoring for operator-initiated disconfirmations comprises receiving further speech signals and calculating, according to a speech recognition algorithm, measures of probability that the further speech signals correspond to disconfirmations of actions selected in response to previously analyzed speech signals.

9. A method as defined in claim 1, wherein the step of calculating, according to a speech recognition algorithm responsive to the representations of speech signals and the further data, comprises calculating, according to a speech recognition algorithm responsive to the representations of speech signals, measures of probability that the speech signals correspond to each of a plurality of actions in a restricted action vocabulary, the restricted action vocabulary being a subset of the action vocabulary selected in response to the further data.

10. A method as defined in claim 1, further comprising using the calculated speech recognition model parameters in subsequent applications of the speech recognition algorithm.

11. A method as defined in claim 1, further comprising an initial step of prompting speakers for speech signals indicative of desired actions.