Methods and apparatus for generating, updating and distributing speech recognition models

• US 6,823,306 B2
• Filed: 11/30/2000
• Issued: 11/23/2004
• Est. Priority Date: 11/30/2000
• Status: Active Grant

• Current Assignee: Google LLC (Alphabet Inc.)
• Original Assignee: Telesector Resources Group Incorporated
• Inventors: Levas, Suzi, Reding, Craig
• Primary Examiner(s): Dorvil, Richemond
• Assistant Examiner(s): Han, Qi
• Application Number:
  US09/726,972
• Publication Number:
  US 20020065656A1
• Time in Patent Office: 1,454 Days
• Field of Search: 704/256, 704/231, 704/275, 704/270.1, 704/235, 704/244
• US Class Current: 704/244
• CPC Class Codes: G10L 15/00 Speech recognition G10L17/00 takes precedence
• CPC Class Codes: G10L 15/063 Training
• CPC Class Codes: G10L 15/30 Distributed recognition, e.g. in client-server systems, for mobile phones or network applications

Abstract

Techniques for generating, distributing, and using speech recognition models are described. A shared speech processing facility is used to support speech recognition for a wide variety of devices with limited capabilities including business computer systems, personal data assistants, etc., which are coupled to the speech processing facility via a communications channel, e.g., the Internet. Devices with audio capture capability record and transmit to the speech processing facility, via the Internet, digitized speech and receive speech processing services, e.g., speech recognition model generation and/or speech recognition services, in response. The Internet is used to return speech recognition models and/or information identifying recognized words or phrases. Thus, the speech processing facility can be used to provide speech recognition capabilities to devices without such capabilities and/or to augment a device'"'"'s speech processing capability. Voice dialing, telephone control and/or other services are provided by the speech processing facility in response to speech recognition results.

Claims

1. A speech processing method, the method comprising the steps of:

transmitting a stored speech recognition model including a first set of speech information to a remote speech processing facility;

transmitting to the remote speech processing facility said stored speech recognition model being used to model a spoken word;

information identifying a said word which is modeled by said speech recognition model;

receiving from the remote speech processing facility a new speech recognition model corresponding to said word, the new speech recognition model including speech characteristic information that is not included in the stored speech recognition model.

2. The method of claim 1, further wherein the step of transmitting the stored speech recognition model includes sending the stored speech recognition model to the remote processing facility via the Internet.

3. The method of claim 2,

wherein said new speech recognition model is a Hidden Markov Model; and

wherein the step of receiving a new speech recognition model includes;

receiving the speech recognition model via the Internet.

4. The method of claim 3, further comprising the steps of:

replacing the stored speech recognition model in a memory device with the received new speech recognition model.

5. The method of claim 4, wherein the stored speech recognition model is a dynamic time warping template and the new speech recognition model is a Hidden Markov Model.

6. The method of claim 1, wherein the speech characteristics information that is not included in the stored speech recognition model includes information regarding changes in signal amplitude as a function of time.

7. A method of generating and distributing speech recognition models, the method comprising the steps of:

receiving speech from a first speech recognition device used by a first user;

generating a first speech recognition model from the first speech;

transmitting the first speech recognition model to the first speech recognition device;

receiving speech from a second speech recognition device, used by a second user the second speech recognition device being different from the first speech recognition device, said first user being different from said second user;

generating a second speech recognition model from the second speech;

transmitting the second speech recognition model to the second speech recognition device; and

transmitting the first and second speech recognition models to a plurality of additional speech recognition devices.

8. The method of claim 7,

wherein the first, second, and plurality of additional speech recognition devices are coupled to the Internet; and

wherein the steps of transmitting to the second and plurality of additional speech recognition devices includes transmitting the second speech recognition model over the Internet.

9. The method of claim 7, further comprising the step of:

storing the first and second speech recognition models in a model store with information indicating when the first and second speech recognition models where created.

10. The method of claim 9, further comprising the step of:

determining when to transmit the first and second speech recognition models as a function of the stored information.