Continuous speech recognition method and program medium with alternative choice selection to confirm individual words

• US 6,564,185 B1
• Filed: 08/10/1999
• Issued: 05/13/2003
• Est. Priority Date: 09/08/1998
• Status: Expired due to Term

• Current Assignee: Seiko Epson Corporation (Seiko Group)
• Original Assignee: Seiko Epson Corporation (Seiko Group)
• Inventors: Inazumi, Mitsuhiro, Ikejiri, Masahisa, Miyazawa, Yasunaga, Hasegawa, Hiroshi
• Primary Examiner(s): Banks-Harold, Marsha D.
• Assistant Examiner(s): Storm, Donald L.
• Application Number:
  US09/370,982
• Time in Patent Office: 1,372 Days
• Field of Search: 704/235, 704/251, 704/271, 704/275
• US Class Current: 704/251
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue

Abstract

The invention relates to a method and apparatus for recognition processing of continuous words of a group which is structured by a plurality of words such that a recognition result of all of the words which structures the continuous words is effectively and accurately confirmed. All of the continuous words which have been input are recognition processed, the recognition result of all of the continuous words is output, a response from a speaker showing an affirmative/negative recognition result is input and recognition processed. If affirmative is determined, the recognition result at that time is confirmed for all of the continuous words. If negative is determined, for each word from a first to an nth (third in this case) which structures continuous words, the content showing affirmative/negative from the speaker is recognized, affirmative or negative is determined, and the recognition result at that time is confirmed as a recognition processing target word.

Claims

1. A continuous word recognition method for use in a speech recognition device, comprising:

recognition processing one or more continuously spoken words input by a speaker;

outputting the recognition processing result to the speaker;

prompting the speaker to confirm the recognition result;

receiving a response from the speaker indicating whether the recognition result is correct;

recognizing the response from the speaker; and

determining whether the response indicates that the recognition result is correct such that if the recognition result is correct, the recognition result is confirmed and stored, and if the recognition result is not correct;

outputting the recognition result word by word from a first to an nth word, n being a positive integer, that form the one or more continuously spoken words;

prompting the speaker to confirm the recognition result of each word;

receiving the speaker'"'"'s confirmation response after each word of the recognition result is output;

determining whether the response indicates that the recognition result of each word is correct such that if the recognition result is correct, the recognition result of each word is confirmed and stored, and if the recognition result for each word is not correct;

prompting the speaker to input another word;

confirming the recognition result for the another input word;

outputting at least one or more predetermined candidates in order, starting with a first candidate, with respect to a word which is a current processing target defined as a recognition target word, among the first to the nth words that form the continuous words;

inputting a response from the speaker indicating an affirmative/negative response per output candidate;

recognition processing the response;

confirming the output candidate as the recognition target word when the response of the speaker is determined to be affirmative;

outputting a following candidate when the response of the speaker is determined to be negative;

inputting the response from the speaker indicating an affirmative/negative response with respect to the output candidate;

recognition processing the affirmative/negative response;

confirming the output candidate as the recognition target word when the response of the speaker is determined to be affirmative;

repeating the outputting the following candidate step and the inputting the response step for the following candidates up to the an mth candidate if a negative response is determined;

outputting a request for the speaker to speak the recognition target word again when the response with respect to the mth candidate is determined to be negative;

outputting one of two words as a recognition result when a word among the first to the nth words that form the continuous words is a word which is mutually exclusive in terms of a meaning; and

confirming the other word of the two words as a recognition result when the response from the speaker indicating an affirmative/negative response with respect to the output is negative.

2. A recording medium for storing programs for performing continuous word recognition, comprising:

a program for recognition processing one or more continuously spoken words input by a speaker;

a program for outputting the recognition processing result to the speaker;

a program for prompting the speaker to confirm the recognition result;

a program for receiving a response from the speaker indicating whether the recognition result is correct;

a program for recognizing the response from the speaker; and

a program for determining whether the response indicates that the recognition result is correct such that if the recognition result is correct, the recognition result is confirmed and stored, and if the recognition result is not correct;

a program for outputting the recognition result word by word from a first to an nth word, n being a positive integer, that form the one or more continuously spoken words;

a program for prompting the speaker to confirm the recognition result of each word;

a program for receiving the speaker'"'"'s confirmation response after each word of the recognition result is output;

a program for determining whether the response indicates that the recognition result of each word is correct such that if the recognition result is correct, the recognition result of each word is confirmed and stored;

a program for prompting the speaker to input another word if the recognition result for each word is not correct;

a program for confirming the recognition result for the another input word;

the program for outputting a following candidate and the program for inputting the response, being repeated for the following candidates up to an mth candidate if a negative response is determined;

a program for outputting one of two words as a recognition result when a word among the first to the nth words that form the continuous words is a word which is mutually exclusive in terms of a meaning; and

a program for confirming the other word of the two words as a recognition result when the response from the speaker indicating an affirmative/negative response with respect to the output is negative.

3. The recording medium for storing programs as set forth in claim 2, further comprising:

a program for outputting at least one or more predetermined candidates in order, starting with a first candidate, with respect to a word which is a current processing target defined as a recognition target word, among the first to the nth of the words that form the continuous words.

4. The recording medium for storing programs as set forth in claim 3, further comprising:

a program for inputting a response from the speaker indicating an affirmative/negative response per output candidate;

a program for recognition processing the response; and

a program for confirming the output candidate as the recognition target word when the response of the speaker is determined to be affirmative.

5. The recording medium for storing programs as set forth in claim 4, further comprising:

a program for outputting a following candidate when the response of the speaker is determined to be negative; and

a program for inputting the response from the speaker indicating an affirmative/negative response with respect to the output candidate;

a program for recognition processing the affirmative/negative, confirming the output candidate as the recognition target word when the response of the speaker is determined to be affirmative.

6. The recording medium for storing programs as set forth in claim 5, the program for outputting the following candidate and the program for inputting the response, being repeated for the following candidates up to the mth candidate if a negative response is determined.

7. The recording medium for storing programs as set forth in claim 6, further comprising:

a program for outputting a request for the speaker to speak the recognition target word again when the response with respect to the mth candidate is determined to be negative.