Multi-user speech recognition system

• US 4,922,538 A
• Filed: 02/05/1988
• Issued: 05/01/1990
• Est. Priority Date: 02/10/1987
• Status: Expired due to Term

• Current Assignee: British Telecommunications PLC (BT Group PLC)
• Original Assignee: British Telecommunications PLC (BT Group PLC)
• Inventors: Tchorzewski, Kazimierz
• Primary Examiner(s): KEMENY, EMANUEL
• Application Number:
  US07/152,924
• Time in Patent Office: 816 Days
• Field of Search: 381/51-53, 381/41-43, 364/513.5
• US Class Current: 704/247
• CPC Class Codes: G10L 15/32 Multiple recognisers used in sequence or in parallel; Score combination systems therefor, e.g. voting systems
• CPC Class Codes: H04M 2201/40 using speech recognition
• CPC Class Codes: H04M 3/42314 in private branch exchanges

Abstract

A multi-user speech recognition system includes several speech recognizers which are shared between many user terminals. When a user terminal is activated, one of the recognizers is assigned to deal with voiced instructions received via that user terminal. This recognizer requests the respective "templates" from a memory device. A controller handles requests from the recognizers on a first-come first-served basis, and prevents later requests from interrupting earlier ones.

Claims

1. A speech recognition system comprising:

a plurality of user terminals connected to a plurality of single-speaker dependent speech recognisers, there being a smaller number of speech recognisers than there are user terminals,

the speech recognisers being arranged such that, when a user terminal is activated, one of the recognisers is assigned to deal with voiced instructions input via that user terminal;

a database containing respective single-speaker dependent templates for each of the user terminals, the database being arranged, in use, to send a copy of one of said templates corresponding to a particular user terminal to a recogniser assigned to deal with instructions input via that user terminal; and

control means connected to the speech recognisers and arranged, in use, to control the transfer of copies of templates from the database to the recognisers;

wherein the control means is arranged, in use, to receive a request for a copy of one of said templates from one of the recognisers, and to instruct the database to send a complete copy of the requested template without interruption by another such request from another terminal.

2. A speech recognition system as claimed in claim 1, wherein the control means comprises a cyclic channel selector and a clock, the cyclic channel selector and the clock being interconnected by a switch which is controlled by the requests for copies of templates from the recognisers to stop the cyclic channel selector for the duration of said transfer of a complete requested template.

3. In a multi-user single-speaker dependent speech recognition system having m speech recognition processors shared by a larger number of n users and having n corresponding speaker dependent speech recognition templates stored in a common data base shared by said m speech recognition processors, the improvement comprising:

means for assigning particular ones of said m speech recognition processors to particular ones of said n users on a first-come first-served basis; and

means for transferring a complete requesting user template to the correspondingly assigned speech recognition processor in response to said assignment and without interruption by any other requesting user.

4. An improved multi-user single-speaker dependent speech recognition system as in claim 3 wherein said means for assigning comprises means for cyclically addressing unassigned ones of said speech recognition processors to effect temporary connection to a common memory device containing said data base.