Speech recognition apparatus, speech recognition method, and speech recognition robot

• US 8,886,534 B2
• Filed: 01/27/2011
• Issued: 11/11/2014
• Est. Priority Date: 01/28/2010
• Status: Active Grant

• Current Assignee: Honda Motor Co., Ltd. (Honda Motor Company)
• Original Assignee: Honda Motor Co., Ltd. (Honda Motor Company)
• Inventors: Nakano, Mikio, Iwahashi, Naoto, Funakoshi, Kotaro, Sumii, Taisuke
• Primary Examiner(s): WOZNIAK, JAMES S
• Application Number:
  US13/015,274
• Publication Number:
  US 20110184737A1
• Time in Patent Office: 1,384 Days
• Field of Search: 704/243, 704/251, 704/254, 704/275, 704/244, 901/1
• US Class Current: 704/243
• CPC Class Codes: G10L 15/12 using dynamic programming techniques, e.g. dynamic time warping [DTW]
• CPC Class Codes: G10L 15/24 Speech recognition using non-acoustical features

Abstract

A speech recognition apparatus includes a speech input unit that receives input speech, a phoneme recognition unit that recognizes phonemes of the input speech and generates a first phoneme sequence representing corrected speech, a matching unit that matches the first phoneme sequence with a second phoneme sequence representing original speech, and a phoneme correcting unit that corrects phonemes of the second phoneme sequence based on the matching result.

Claims

1. A speech recognition apparatus comprising

a storage medium storing a relationship between a recognition reliability and a correct answer rate for every phoneme;

and

a processor programmed with instructions that, when executed, cause the processor to;

receive speech;

recognize phonemes of the speech and generate a first phoneme sequence representing corrected speech;

match the first phoneme sequence with a second phoneme sequence representing original speech so as to create a pair of a first phoneme extracted from the first phoneme sequence and a second phoneme extracted from the second phoneme sequence; and

replace the second phoneme with the first phoneme when the second phoneme differs from the first phoneme, a predetermined recognition reliability of the first phoneme is higher than a predetermined recognition reliability of the second phoneme, and a correct answer rate of the first phoneme determined based on the relationship stored in the storage medium is higher than a predetermined value.

2. The speech recognition apparatus according to claim 1, wherein the processor is further programmed with instructions that, when executed, cause the processor to calculate distances between the phonemes of the first phoneme sequence and the phonemes of the second phoneme sequence based on a frequency of each set of phoneme types recognized as the types of phonemes included in the input speech and to determine the matching result based on the distances.

3. The speech recognition apparatus according to claim 1, further comprising a dialogue response pattern storage medium storing response patterns representing the recognition result, wherein

the processor is further programmed with instructions that, when executed, cause the processor to:

reproduce speech based on the corrected second phoneme sequence;

and

perform one of a process of storing word information including the corrected second phoneme sequence in a word storage medium and a process of reproducing speech urging a user to utter speech based on the response pattern matched with the phonemes of the input speech.

4. A speech recognition robot comprising the speech recognition apparatus according to claim 1.

5. A speech recognition method, comprising:

storing, in a storage medium, a relationship between a recognition reliability and a correct answer rate for every phoneme;

receiving speech;

recognizing phonemes of the speech;

generating a first phoneme sequence representing corrected speech;

matching the first phoneme sequence with a second phoneme sequence representing original speech so as to create a pair of a first phoneme extracted from the first phoneme sequence and a second phoneme extracted from the second phoneme sequence; and

replacing the second phoneme with the first phoneme when the second phoneme differs from the first phoneme, a predetermined recognition reliability of the first phoneme is higher than a predetermined recognition reliability of the second phoneme, and a correct answer rate of the first phoneme determined based on the relationship stored in the storage medium is higher than a predetermined value.