Method and a system for substantially eliminating speech recognition error in detecting repetitive sound elements

• US 6,157,911 A
• Filed: 03/27/1998
• Issued: 12/05/2000
• Est. Priority Date: 03/28/1997
• Status: Expired due to Term

• Current Assignee: Ricoh Company Limited
• Original Assignee: Ricoh Company Limited
• Inventors: Kuroda, Masaru
• Primary Examiner(s): Hudspeth, David R.
• Assistant Examiner(s): Azad, Abul K.
• Application Number:
  US09/049,371
• Time in Patent Office: 984 Days
• Field of Search: 704/251, 704/252, 704/253, 704/254, 704/255, 704/231, 704/239, 704/241
• US Class Current: 704/251
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue
• CPC Class Codes: G10L 2015/088 Word spotting

Abstract

A method and a system substantially eliminates an erroneous voice recognition of repetitive elements in word spotting. One preferred embodiment according to the current invention eliminates erroneous voice recognition of repetitive elements by selectively prolonging a response time of words containing repetitive elements. In order to substantially eliminate the errors, in another preferred embodiment according to the current invention, words containing repetitive elements are marked by a silent key word.

Claims

1. A method of substantially eliminating an error in speech recognition, comprising the steps of:

a) modifying a response time for an entry having repetitive voice elements in a predetermined dictionary;

b) inputting voice data;

c) performing word spotting for matching a portion of the voice data with the entry of the predetermined dictionary so as to determine a similarity degree; and

d) delaying to output a similarity degree upon detecting the repeated voice elements, wherein said word spotting substantially eliminating an erroneous additional detection of the repeated voice elements.

2. The method of substantially eliminating an error in speech recognition according to claim 1 wherein said response time is allocated for 0.2 second at the end of the entry.

3. The method of substantially eliminating an error in speech, recognition according to claim 1 wherein said response time is prolonged by 0.3 second for the entry having repetitive voice elements.

5. The method of substantially eliminating an error in speech recognition according to claim 1 wherein said silent key is a predetermined letter.

4. A method of substantially eliminating an error speech recognition, comprising the steps of:

e) adding a silent key at an end of an entry having repeated voice elements in a predetermined dictionary;

f) inputting voice data; and

g) performing word spotting for matching a portion of the voice data with the entry of the predetermined dictionary, wherein upon detecting the repeated voice elements, said word spotting substantially eliminating an erroneous additional detection of the repeated voice elements.

6. A system for substantially eliminating an error in speech recognition, comprising:

a standard dictionary having a prolonged response time for an entry having repeated voice elements in a predetermined dictionary;

an input unit for inputting voice data; and

an extraction characteristic unit connected to said standard dictionary for performing word spotting for matching a portion of the voice data with the entry of the predetermined dictionary and outputting a similarity degree, wherein upon detecting the repeated voice elements, said word spotting substantially eliminating an erroneous additional detection of the repeated voice elements based upon the prolonged response time prior to outputting the similarity degree.

7. The method of substantially eliminating an error in speech recognition according to claim 6 wherein said prolonged response time is approximately 0.5 second.

8. A system for substantially eliminating an error in speech recognition, comprising:

a standard dictionary having a silent key for an entry having repeated voice elements in a predetermined dictionary;

an input unit for inputting voice data; and

an extraction characteristic unit connected to said standard dictionary for performing word spotting for matching a portion of the voice data with the entry of the predetermined dictionary and outputting a similarity degree, wherein upon detecting the repeated voice elements, said word spotting substantially eliminating an erroneous additional detection of the repeated voice elements based upon the silent key.

9. The system for substantially eliminating an error in speech, recognition according to claim 8 wherein said silent key is a predetermined letter.