Speech recognizing method and apparatus, and speech translating system

• US 5,848,389 A
• Filed: 04/05/1996
• Issued: 12/08/1998
• Est. Priority Date: 04/07/1995
• Status: Expired due to Term

• Current Assignee: Sony Corporation (Sony Group Corp.)
• Original Assignee: Sony Corporation (Sony Group Corp.)
• Inventors: Asano, Yasuharu, Ogawa, Hiroaki, Kagami, Tetsuya, Akabane, Makoto, Ishii, Kazuo, Kakuda, Hiroshi, Tanaka, Miyuki, Kato, Yasuhiko, Watari, Masao
• Primary Examiner(s): Dorvil, Richemond
• Application Number:
  US08/626,132
• Time in Patent Office: 977 Days
• Field of Search: 395/2.48, 395/2.86, 395/2.4, 395/2.49, 395/2.6, 395/2.61, 395/2.64, 704/239, 704/277, 704/243, 704/231, 704/240, 704/251, 704/252, 704/255, 704/2, 704/4, 704/7, 704/9
• US Class Current: 704/239
• CPC Class Codes: G10L 15/18 using natural language modelling
• CPC Class Codes: G10L 15/19 Grammatical context, e.g. disambiguation of the recognition hypotheses based on word sequence rules
• CPC Class Codes: G10L 2015/085 Methods for reducing search complexity, pruning

Abstract

In a speech recognizing apparatus, a grammatical qualification of a proposed speech recognition result candidate is judged without using a grammatical rule. The speech recognizing apparatus for performing sentence/speech recognition is comprised of an analyzing unit for acoustically analyzing speech inputted therein to extract a feature parameter of the inputted speech; a recognizing unit for recognizing the inputted speech based upon the feature parameter outputted from said analyzing unit to thereby a plurality of proposed recognition result candidates; an example data base for storing therein a plurality of examples; and an example retrieving unit for calculating a resemblance degree between each of said plurality of proposed recognition result candidates and each of the plural examples stored in the example data base and for obtaining the speech recognition result based on said calculated resemblance degree.

Claims

1. A speech recognizing apparatus for performing sentence/speech recognition, comprising:

extracting means for acoustically analyzing speech inputted therein to extract a feature parameter of said inputted speech;

recognizing means for recognizing said inputted speech based upon said feature parameter outputted from said extracting means to thereby obtain a plurality of proposed recognition result candidates;

example memory means for storing therein a plurality of examples; and

calculation means for calculating a resemblance degree between each of said plurality of proposed recognition result candidates and each of the plural examples stored in said example memory means and for obtaining said speech recognition result based on said calculated resemblance degree.

2. A speech recognizing apparatus as claimed in claim 1 wherein:

said recognizing means recognizes said input speech in word units.

3. A speech recognizing apparatus as claimed in claim 2, further comprising:

word concept memory means for storing therein words by categorizing said words;

wherein

each proposed recognition result candidate is made by combining more than one word; and

said calculating means calculates a word resemblance degree indicative of a conceptional resemblance characteristic between each of said words for constituting said proposed recognition result candidates and each of the words for constituting said examples with reference to said word concept memory means, and calculates said resemblance degree based on said calculated word resemblance degree.

4. A speech recognizing apparatus as claimed in claim 3, wherein:

said word concept memory means stores therein the words in a hierarchical tree structure based on the concept thereof.

5. A speech recognizing apparatus as claimed in claim 3, further comprising:

word memory means for storing therein at least the word to be recognized by said recognizing means, phonemic information about said word, and speech part information about said word;

wherein

said word concept memory means and said word memory means are made in an integral form.

6. A speech recognizing apparatus as claimed in claim 1, further comprising:

language model memory means for storing therein a language model classified according to the probability of connections among a plurality of words;

wherein

said recognizing means performs said speech recognition under restrictions of said language model stored in said language model storing means so as to obtain a plurality of proposed recognition result candidates.

7. A speech recognizing method for performing sentence/speech recognition, comprising the steps of:

acoustically analyzing input speech to extract a feature parameter of said input speech;

recognizing said input speech based upon said extracted feature parameter so as to obtain a plurality of proposed recognition result candidates; and

calculating a resemblance degree between each of said plurality of proposed recognition result candidates and each of plural examples stored in example memory means and then obtaining said speech recognition result based upon said calculated resemblance degree.

8. A speech translating apparatus for translating a language of inputted speech into another language, comprising:

extracting means for acoustically analyzing speech inputted therein to extract a feature parameter of said inputted speech;

recognizing means for recognizing said inputted speech based upon said feature parameter outputted from said extracting means to thereby obtain a plurality of proposed recognition result candidates;

example memory means for storing therein a plurality of examples;

calculation means for calculating a resemblance degree between each of said plurality of proposed recognition result candidates and each of the plural examples stored in said example memory means and for obtaining said speech recognition result based on said calculated resemblance degree, and

translating means for translating the language of said speech recognition result outputted from said calculation means into said another language.

9. A speech translating apparatus as claimed in claim 8 wherein

said recognizing means recognizes said input speech in word units.

10. A speech translating apparatus as claimed in claim 8, further comprising:

word concept memory means for storing therein words by categorizing said words;

wherein

each proposed recognition result candidate is made by combining more than one word; and

said calculating means calculates a word resemblance degree indicative of a conceptional resemblance characteristic between each of said words for constituting said proposed recognition result candidates and each of the words for constituting said examples with reference to said word conceptional memory means, and calculates said resemblance degree based on said calculated word resemblance degree.

11. A speech translating apparatus as claimed in claim 10, wherein

said word concept memory means stores therein the words in a hierarchical tree structure based on the concept thereof.

12. A speech translating apparatus as claimed in claim 10, further comprising:

word memory means for storing therein at least the word to be recognized by said recognizing means, phonemic information about said word, and part of speech information about said word;

wherein

said word concept memory means and said word memory means are made in an integral form.

13. A speech translating apparatus as claimed in claim 8, further comprising:

language model memory means for storing therein a language model classified according to the probability of connections among a plurality of words;

wherein;

said recognizing means performs said speech recognition under restriction of said language model stored in said language model storing means so as to obtain a plurality of proposed recognition result candidates.