Data processing: speech signal processing – linguistics – language – Speech signal processing – Recognition
Reexamination Certificate
2000-09-29
2004-12-28
Dorvil, Richemond (Department: 2641)
Data processing: speech signal processing, linguistics, language
Speech signal processing
Recognition
C704S238000, C704S245000, C704S252000
Reexamination Certificate
active
06836760
ABSTRACT:
BACKGROUND OF THE INVENTION
Field of the Invention
This invention relates, in one embodiment, generally to speech recognition systems, and more particularly to using semantic inference with speech recognition systems.
COPYRIGHT NOTICE/PERMISSION
A portion of the disclosure of this patent document contains material that is subject to copyright protection. The copyright owner has no objection to the facsimile reproduction by anyone of the patent document or the patent disclosure as it appears in the Patent and Trademark Office patent file or records, but otherwise reserves all copyright rights whatsoever. The following notice applies to the software and data as described below and in the drawings hereto: Copyright © 2000, Apple Computer, Inc., All Rights Reserved.
BACKGROUND
Speech recognition enables a computer system to receive voice inputs and convert them into text. The computer system receives an audio input, transforms the audio input into digital data, compares the digital data with a list of digitized waveforms corresponding to text, and converts the digital data into the text corresponding to the most closely matched digitized waveform. One application of speech recognition is voice command and control (VCC), which enables a computer user to control a computer by voice rather than by using traditional user interfaces such as a keyboard and a mouse. Advances in speech recognition technology have enhanced the performance of VCC so that a computer can accurately perform a task by recognizing a command spoken within a restricted domain of vocabulary. However, existing VCC technology has limitations that diminish the usefulness of the technology to an average computer user.
Typical VCC applications employ a context-free grammar, such as a finite state grammar, that is a compact way of representing an exhaustive list of each and every command that the application can recognize. A finite state grammar is a particular implementation of a context-free grammar. These applications compare the spoken command to the list of commands underlying the context-free grammar. Previously existing VCC applications that use a context-free grammar either reject or incorrectly recognize any utterance that is semantically accurate but syntactically out-of-grammar. This rigid framework requires the computer user to learn and memorize the specific commands that are compiled within the context-free grammar.
Semantic inference alleviates the problems associated with VCC applications that use a context-free grammar. Semantic inference is a more tolerant approach to language modeling that enables a computer to recognize commands that are out-of-grammar but semantically accurate, thereby allowing computer users to say what they mean rather than requiring them to speak from an established list of commands. Existing semantic inference systems replace a context-free grammar in a speech recognition unit with a statistical language model such as an n-gram. This substitution prevents the speech recognition unit from rejecting out-of-grammar voice inputs before the semantic classification engine has the opportunity to evaluate the voice input for semantic similarity. A statistical language model makes it possible for the speech recognition unit to transcribe, with a reasonably low error rate, whatever formulation the computer user chooses for expressing a command. A semantic classification engine then operates on the transcription to determine the desired action.
Using a statistical language model with the speech recognition unit enables the voice command and control system to accurately identify the correct command. However, there are problems associated with semantic inference systems that employ a statistical language model. Substituting a statistical language model for a context-free grammar in the speech recognition unit requires a significant change in the overall architecture of the speech recognition unit, specifically in the structure of the search module. Also, estimating the parameters of the statistical language model typically requires multiple iterations over a large training corpus of relevant text data, which may involve a large number of central processor unit (CPU) cycles. Additionally, developing and maintaining such a large corpus of text data is time-consuming and expensive. Furthermore, a speech recognition unit using a statistical language model typically requires the computer user to wear a head-mounted noise-canceling microphone and to train the system to his or her voice. Finally, n-gram statistical language models have significantly larger storage requirements than context-free grammars and lead to greater recognition runtimes.
Therefore, a method and apparatus to use semantic inference with a speech recognition system using a context-free grammar are required.
SUMMARY OF THE INVENTION
A method and apparatus to use semantic inference with speech recognition systems using a context-free grammar is described herein. According to one aspect of the invention, a method for speech recognition comprises recognizing at least one spoken word, processing the spoken word using a context-free grammar, deriving an output from the context-free grammar, and translating the output into a predetermined command.
According to one aspect of the present invention, a machine-readable medium has stored thereon a plurality of instructions that, when executed by a processor, cause the processor to recognize at least one spoken word, process the spoken word using a context-free grammar, derive an output from said context-free grammar, and translate the output into a predetermined command.
According to one aspect of the present invention, an apparatus for speech recognition includes a processing unit, a memory unit, a system bus, and at least one machine-readable medium. A speech recognition unit, a context-free grammar, and a semantic inference engine are stored in the machine-readable medium.
REFERENCES:
patent: 4974191 (1990-11-01), Amirghodsi et al.
patent: 5642519 (1997-06-01), Martin
patent: 5652897 (1997-07-01), Linebarger et al.
patent: 5774859 (1998-06-01), Houser et al.
patent: 6023676 (2000-02-01), Erell
Jerome R. Bellagarda, J. W. Butzberger, Y-L Chow, N. B. Coccaro, and D. Naik, “A Novel Word Clustering Algorithm Based On Latent Semantic Analysis,” IEEE Int. Conf. on Acoust., Speech, and Sig. Proc. 1996 ICASSP-96, May 7-10, 1996, vol. 1, pp. 172-175.*
F. Jelinek, “Self-Organized Language Modeling for Speech Recognition,” in Reading in Speech Recognition, A. Waibel and K.F Lee, Eds, Morgan Kaufmann Publishers, pp. 450-506, 1990.
J.R. Bellegarda and K.E.A. Silverman, Data-Driven Semantic Inference for Unconstrained Command and Control by Voice, U.S. patent application Ser. No.: 09/183,070, filed Oct. 1998.
F. Kubala et al., “The Hub and Spoke Paradigm for CSR Evaluation”, in Proc. ARPA Spoken Language Technology Workshop, Morgan Kaufmann Publishers, 6 pages, Mar. 1994.
Deerwester, S. et al. “Indexing By Latent Semantic Analysis,” Journal of the American Society for information Science, vol. 41, No. 6, Sep. 1990, pp 391-407.
J. Caminero-Gil, et al., “Data-Driven Disclosure Modeling for Semantic Interpretation,” Proc. IEEE ICASSP 96, vol. 1,p. 401-404.
Jerome R. Bellegarda, “A Multispan Language Modeling Framework for large Vocabulary Speech Recognition,” IEEE Trans. Speech and Audio Processing, vol. 6, No. 5, p. 456-467.
Bellegarda Jerome R.
Silverman Kim E. A.
Apple Computer Inc.
Blakely , Sokoloff, Taylor & Zafman LLP
Dorvil Richemond
Storm Donald L.
LandOfFree
Use of semantic inference and context-free grammar with... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Use of semantic inference and context-free grammar with..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Use of semantic inference and context-free grammar with... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3312121