System and method for modeless large vocabulary speech...

Data processing: speech signal processing – linguistics – language – Speech signal processing – Recognition

Reexamination Certificate

Rate now

  [ 0.00 ] – not rated yet Voters 0   Comments 0

Details

C704S256000

Reexamination Certificate

active

06292779

ABSTRACT:

FIELD OF THE INVENTION
The present invention relates to a speech recognition system, and more particularly to a flexible speech recognition system for large vocabulary continuous speech dictation which also recognizes and acts upon command and control phrases embedded in a user provided dictation stream.
BACKGROUND ART
Speech recognition systems allow a user to operate and control other applications such as word processors, spreadsheets, databases, etc. Accordingly, a useful speech recognition system allows a user to perform to broad functions: (1) dictate input to an application, and (2) control the input and the application. One approach of prior art systems has been to provide separate dictation processing and control processing modes and require the user to switch between the two modes. Thus, operating mode would be definitely known by the system, since positive direction by the user was necessary to change processing modes.
Another approach was described by Hsu in U.S. Pat. No. 5,677,991 and Yegnanarayanan in U.S. Pat. No. 5,794,196, both of which are incorporated herein by reference in their entirety, in which input speech was parsed by both a large vocabulary isolated word recognition module and a small vocabulary continuous speech recognition module each having an associated application context. Hypotheses produced by the large vocabulary isolated word speech recognition module would correspond to dictated text while hypotheses produced by the small vocabulary continuous speech recognition module would correspond to short application specific command and control sequences. Each recognition module would produce hypotheses corresponding to the input speech and an associated recognition probability or score. An arbitration algorithm would then select the better scoring hypothesis as a recognition result and direct the result to the associated context.
The approach of Hsu and Yegnanarayanan represented an advance in that a user of the speech recognition system no longer needed to toggle between dictation mode and command mode, rather the system automatically determined whether a given portion of an input utterance should be treated as dictated text or as application related command directives. However, Hsu and Yegnanarayanan explicitly limit the large vocabulary speech recognition module to an isolated word approach which requires a user to pause unnaturally between each word of dictated text.
SUMMARY OF THE INVENTION
A preferred embodiment of the present invention represents a method for operating a modeless large vocabulary continuous speech recognition system of the type that represents an input utterance as a sequence of input vectors. The method includes:
(a) providing a common library of acoustic model states for arrangement in sequences that form acoustic models;
(b) comparing each vector in a sequence of input vectors to a set of model states in the common library to produce a match score for each model state in the set reflecting the likelihood that such state is represented by such vector; and
(c) using, in a plurality of recognition modules operating in parallel, the match scores with the acoustic models to determine at least one recognition result in each of the recognition modules.
In a further embodiment, each acoustic model may be composed of a sequence of segment models and each segment model may be composed of a sequence of model states. The match score may be a probability calculation or a distance measure calculation. Each recognition module may include a recognition grammar used with the acoustic models to determine the at least one recognition result. The recognition grammar may be a context-free grammar, a natural language grammar, or a dynamic command grammar. In addition, or alternatively, the method may further include comparing the recognition results of the recognition modules to select at least one system recognition result. The step of comparing may use an arbitration algorithm and a score ordered queue of recognition results and associated recognition modules. The plurality of recognition modules may include one or more of a dictation module for producing at least one probable dictation recognition result, a select module for recognizing a portion of visually displayed text for processing with a command, and a command module for producing at least one probable command recognition result.
A related embodiment provides a modeless large vocabulary continuous speech recognition system of the type that represents an input utterance as a sequence of input vectors. The system includes a common library of acoustic model states for arrangement in sequences that form acoustic models; an input processor that compares each vector in a sequence of input vectors to a set of model states in the common library to produce a match score for each model state in the set reflecting the likelihood that such state is represented by such vector; and a plurality of recognition modules operating in parallel that use the match scores with the acoustic models to determine at least one recognition result in each of the recognition modules.
In a further embodiment, each acoustic model may be composed of a sequence of segment models and each segment model may be composed of a sequence of model states. The match score may be a probability calculation or a distance measure calculation. Each recognition module may include a recognition grammar used with the acoustic models to determine the at least one recognition result. The recognition grammar may be a context-free grammar, a natural language grammar, or a dynamic command grammar. In addition, or alternatively, the system may further include an arbitrator that compares the recognition results of the recognition modules to select at least one system recognition result. The arbitrator may include an arbitration algorithm and a score ordered queue of recognition results and associated recognition modules. The plurality of recognition modules may include one or more of a dictation module for producing at least one probable dictation recognition result, a select module for recognizing a portion of visually displayed text for processing with a command, and a command module for producing at least one probable command recognition result.


REFERENCES:
patent: 5677991 (1997-10-01), Hsu et al.
patent: 5737489 (1998-04-01), Chou et al.
patent: 5794196 (1998-08-01), Yegnanarrayanan et al.
patent: 5799279 (1998-08-01), Gould et al.
patent: 5832430 (1998-11-01), Lleida et al.
patent: 5850627 (1998-12-01), Gould et al.
patent: 6029124 (2000-02-01), Gillick et al.
patent: 6076056 (2000-06-01), Huang et al.
patent: 196 35 754 (1996-03-01), None
patent: WO 96/13829 (1996-09-01), None

LandOfFree

Say what you really think

Search LandOfFree.com for the USA inventors and patents. Rate them and share your experience with other people.

Rating

System and method for modeless large vocabulary speech... does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with System and method for modeless large vocabulary speech..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and System and method for modeless large vocabulary speech... will most certainly appreciate the feedback.

Rate now

     

Profile ID: LFUS-PAI-O-2524524

  Search
All data on this website is collected from public sources. Our data reflects the most accurate information available at the time of publication.