Speech recognition system having parallel large vocabulary recognition engines

• US 6,526,380 B1
• Filed: 08/09/1999
• Issued: 02/25/2003
• Est. Priority Date: 03/26/1999
• Status: Expired due to Term

• Current Assignee: Huawei Technologies Co., Ltd. (Huawei Investment & Holding Co., Ltd.)
• Original Assignee: Koninklijke Philips Electronics N.V. (Koninklijke Philips N.V.)
• Inventors: Thelen, Eric, Besling, Stefan, Ullrich, Meinhard
• Primary Examiner(s): MCFADDEN, SUSAN IRIS
• Application Number:
  US09/370,490
• Time in Patent Office: 1,296 Days
• Field of Search: 704/231, 704/243, 704/250, 704/251, 704/255, 704/270, 704/256
• US Class Current: 704/251
• CPC Class Codes: G10L 15/18 using natural language modelling
• 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: G10L 2015/228 of application context

Abstract

A huge vocabulary speech recognition system for recognizing a sequence of spoken words, having an input means for receiving a time-sequential input pattern representative of the sequence of spoken words. The system further includes a plurality of large vocabulary speech recognizers each being associated with a respective, different large vocabulary recognition model. Each of the recognition models is targeted to a specific part of the huge vocabulary. The system comprises a controller operative to direct the input pattern to a plurality of the speech recognizers and to select a recognized word sequence from the word sequences recognized by the plurality of speech recognizers.

Claims

1. A huge vocabulary speech recognition system for recognizing a sequence of spoken words, the system comprising

input means for receiving a time-sequential input pattern representative of the sequence of spoken words;

and

a large vocabulary speech recognizer operative to recognize the input pattern as a sequence of words from the vocabulary using a large vocabulary recognition model associated with the speech recognizer;

characterized in that

the system comprises a plurality of N large vocabulary speech recognizers, each being associated with a respective, different large vocabulary recognition model;

each of the recognition models being targeted to a specific part of the huge vocabulary; and

the system comprises a controller operative to direct the input pattern to a plurality of the speech recognizers and to select a recognized word sequence from the word sequences recognized by the plurality of speech recognizers.

2. A system as claimed in claim 1, wherein the system comprises M large vocabulary recognition models, M>

N, and the system comprises a model selector operative to select for at least one of the speech recognizers the associated recognition model from the M models in dependence on a recognition context.

3. A system as claimed in claim 2, wherein the sequence of spoken words relates to a document that determines at least one recognition context.

4. A system as claimed in claim 3, wherein the document is a Web page, such as an HTML page, and the context(s) of the document is/are specified in the document or in association with the document.

5. A system as claimed in claim 3, wherein the model selector is operative to select the recognition model in dependence on words in or associated with the document.

6. A system as claimed in claim 2, wherein the model selector is operative to:

select a test recognition model from the N-M recognition models not yet used by one of the recognizers;

control a testing recognizer to recognize at least part of the input pattern with the test recognition model; and

if the recognition result of the testing recognizer is better than the recognition result of one of the recognizers, enable recognition with the test recognition model.

7. A system as claimed in claim 1, wherein the recognition models are hierarchically arranged from models with a more generic context to models with a more specific context, and wherein the model selector is operative to enable recognition with a more specific model if recognition with the hierarchically-related more generic model at a higher level in the hierarchy obtains good recognition results compared to results of at least one recognizer associated with another recognition model.

8. A system as claimed in claim 1, wherein the system comprises a user station and a server station connected via a network, such as Internet;

the user station being operative to receive the input pattern from a user and transfer a signal representative of the input pattern to the server station;

the server station comprising the recognizers and controller.