SELECTIVE ENABLEMENT OF SPEECH RECOGNITION GRAMMARS

• US 20080189111A1
• Filed: 03/05/2008
• Published: 08/07/2008
• Est. Priority Date: 06/15/2001
• Status: Active Grant

• Current Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Original Assignee: International Business Machines Corporation
• Inventors: Woodward, Steven, Ruback, Harvey
• Granted Patent: US 7,610,204 B2
• Time in Patent Office: Days
• US Class Current: 704/257
• CPC Class Codes: G10L 15/19 Grammatical context, e.g. disambiguation of the recognition hypotheses based on word sequence rules
• CPC Class Codes: G10L 15/30 Distributed recognition, e.g. in client-server systems, for mobile phones or network applications

Abstract

A method for processing speech audio in a network connected client device can include selecting a speech grammar for use in a speech recognition system in the network connected client device; characterizing the selected speech grammar; and, based on the characterization, determining whether to process the speech grammar locally in the network connected client device, or remotely in a speech server in the network. In one aspect of the invention, the selecting step can include establishing a communications session with a speech server; and, querying the speech server for a speech grammar over the established communications session. Additionally, the selecting step can further include registering the speech grammar in the speech recognition system.

Claims

1. A computer-readable storage, having stored thereon a computer program for processing speech audio in a network connected client device, said computer program having a plurality of code sections executable by said client device for causing the client device to perform the steps of:

selecting a speech grammar for use in a speech recognition system in the network connected client device;

characterizing said selected speech grammar, wherein said characterization comprises determining a size and a complexity of said selected grammar and a preferred processing location is specified in said selected grammar;

determining a processing power of said client device and of a remote speech server, a speed of a network connection between said client device and said speech server, and a feedback requirement for said speech recognition system; and

,

based on the characterization of said selected speech grammar, said determined network connection speed, said determined processing power of the network connected client device and the remote speech server, and said feedback requirements, electing whether to process the entire selected speech grammar in said preferred location or another location different from said preferred location before processing the speech audio, wherein said preferred location specifies the network connected client device or the speech server,

wherein if said preferred location specifies said speech server, said client device elects said client device if real-time feedback is required by said speech recognition system and a processing power of said client device is sufficient for said client device to process said selected grammar in real-time based on said size and said complexity of said selected grammar, and wherein if said preferred location specifies said client device, said client device elects said remote speech server if a latency in processing said selected speech grammar based on said network speed and said remote speech server processing power is sufficient to meet a feedback requirement of said speech recognition system.

2. The computer-readable storage of claim 1, wherein the selecting step comprises:

establishing a communications session with said remote speech server; and

,

querying said remote speech server for a speech grammar over said established communications session.

3. The computer-readable storage of claim 1, wherein the selecting step comprises:

establishing a communications session with said remote speech server;

selecting a speech grammar stored in the network connected client device; and

,

uploading the selected speech grammar to said remote speech server.

4. The computer-readable storage of claim 2, wherein said selecting step further comprises:

registering said selected speech grammar in said speech recognition system.

5. The computer-readable storage of claim 2, wherein said characterizing step comprises:

identifying in said selected speech grammar an embedded pre-determined characterization of said size and said complexity of said selected speech grammar.

6. A system for processing speech audio comprising:

a speech processing server; and

a client device for operating a speech recognition system, wherein said client device is communicatively linked to said speech server using a network connection, wherein said client device is operable to;

select a speech grammar for use in the speech recognition system,

characterize the selected speech grammar by determining a size and a complexity of said selected grammar and a preferred processing location is specified in said selected grammar;

determine a processing power of said client device and of said speech processing server, a speed of said network connection, and a feedback requirement for said speech recognition system, and

based on the characterization of the selected speech grammar, said determined network connection speed, said determined processing power of the client device and the remote speech server, and said feedback requirements, elect whether to process the entire selected speech grammar in said preferred location or another location different from said preferred location before processing the speech audio, wherein said preferred location specifies the client device or the speech processing server,

wherein if said preferred location specifies said speech server, said client device elects said client device if real-time feedback is required by said speech recognition system and a processing power of said client device is sufficient for said client device to process said selected grammar in real-time based on said size and said complexity of said selected grammar, and wherein if said preferred location specifies said client device, said client device elects said remote speech server if a latency in processing said selected speech grammar based on said network speed and said speech server processing power is sufficient to meet a feedback requirement of said speech recognition system.

7. The system of claim 6, wherein said speech processing server further comprises a mass storage element for storing a plurality of grammars, and wherein said client device is further operable to:

establish a communications session with said speech processing server; and

,

query said speech processing server for a speech grammar over said established communications session.

8. The system of claim 6, wherein said client device server further comprises a mass storage element for storing said selected grammar, and wherein said client device is further operable to:

establish a communications session with said speech processing server, and

upload the selected speech grammar to said speech processing server.

9. The method of claim 6, wherein said client device is further operable to:

register said selected speech grammar in said speech recognition system.

10. The method of claim 6, wherein said client device is further operable to characterize said size and said complexity of said selected speech by extracting from said selected speech grammar an embedded pre-determined characterization of said size and complexity.