Language processing system

• US 7,917,352 B2
• Filed: 08/24/2006
• Issued: 03/29/2011
• Est. Priority Date: 08/24/2005
• Status: Expired due to Fees

• Current Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation)
• Original Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation)
• Inventors: Seto, Shigenobu
• Primary Examiner(s): Hudspeth; David R
• Assistant Examiner(s): Spooner; Lamont M
• Application Number:
  US11/508,841
• Publication Number:
  US 20070055496A1
• Time in Patent Office: 1,678 Days
• Field of Search: 704/9, 704/258, 704/260, 704 2- 8, 715/264
• US Class Current: 704/8
• CPC Class Codes: G06F 40/284 Lexical analysis, e.g. tokenisation or collocates
• CPC Class Codes: G06F 40/53 Processing of non-Latin text kana-to-kanji conversion G06F40/129; vowelisation G06F40/232
• CPC Class Codes: G10L 13/027 Concept to speech synthesisers; Generation of natural phrases from machine-based concepts generation of parameters for speech synthesis out of text G10L13/08

Abstract

A language processing system including: a forbidden word memory part that stores a forbidden word; a sequence candidate generator that generates a plurality of word sequence candidates where each words are described separately from plain text; and a word sequence estimator that reads the forbidden word from the forbidden word memory part, excludes the word sequence candidate containing the forbidden word from the plurality of word sequence candidates, and selects an estimated word sequence with the highest concatenation possibility of the words from among the plurality of word sequence candidates.

Claims

1. A language processing system comprising:

an input unit that receives plain text including a plurality of input characters in which no separators are included therebetween;

a forbidden word memory part that stores a predetermined forbidden word pronunciation in association with a word having one or more characters;

a sequence candidate generator that identifies separate words and associated pronunciations from the plurality of input characters by referring to a dictionary, generates a plurality of word sequence candidates including a plurality of the separate identified word pronunciations, and stores the plurality of word sequence candidates in a word sequence memory part, wherein each of the word sequence candidates includes all of the input characters but has a different identified word pronunciation than the other word sequence candidates; and

a word sequence estimator that reads the predetermined forbidden word pronunciation from the forbidden word memory part and the plurality of word sequence candidates from the word sequence memory part, determines whether each of the plurality of word sequence candidates contains the predetermined forbidden word pronunciation, eliminates each of the plurality of word sequence candidates containing the predetermined forbidden word pronunciation, and selects an estimated word sequence with the highest concatenation possibility of the words from among the plurality of remaining word sequence candidates.

2. The language processing system according to claim 1, further comprising:

a mis-pronounced range instruction part that accepts instruction of a mis-pronounced word with different reading from a correct reading of text in the estimated word sequence.

3. The language processing system according to claim 2, further comprising:

a forbidden word addition section that additionally stores the mis-pronounced word in the forbidden word memory part as the forbidden word pronunciation.

4. The language processing system according to claim 1, further comprising:

an instructed word estimator that performs a comparison between a reading of the estimated word sequence and a correct reading of text in the estimated word sequence, and extracts a mis-pronounced word with different reading from the correct reading from the estimated word sequence.

5. The language processing system according to claim 4, further comprising:

a forbidden word addition section that additionally stores the mis-pronounced word in the forbidden word memory part as the forbidden word pronunciation.

6. A language processing system comprising:

an input unit that receives plain text including a plurality of input characters in which no separators are included therebetween;

a forbidden word memory part that stores a predetermined forbidden word pronunciation in association with a word having one or more characters;

a sequence candidate generator that reads the predetermined forbidden word pronunciation stored in the forbidden word memory part, identifies separate words and associated pronunciations from the plurality of input characters by referring to a dictionary, determines whether each of the plurality of identified word pronunciations match the predetermined forbidden word pronunciation, excludes each of the plurality of word pronunciations matching the predetermined forbidden word pronunciation, generates a plurality of word sequence candidates including a plurality of the separate identified word pronunciations, and stores the plurality of word sequence candidates in a word sequence memory part, wherein each of the word sequence candidates includes all of the input characters but has a different identified word pronunciation than the other word sequence candidates; and

a word sequence estimator that selects an estimated word sequence with the highest concatenation possibility of the words from among the plurality of word sequence candidates.

7. The language processing system according to claim 6, further comprising:

a mis-pronounced range instruction part that accepts instruction of a mis-pronounced word with different reading from a correct reading of text in the estimated word sequence.

8. The language processing system according to claim 7, further comprising:

a forbidden word addition section that additionally stores the mis-pronounced word in the forbidden word memory part as the forbidden word pronunciation.

9. The language processing system according to claim 6, further comprising:

an instructed word estimator that performs a comparison between a reading of the estimated word sequence and a correct reading of text in the estimated word sequence, and extracts a mis-pronounced word with different reading from the correct reading from the estimated word sequence.

10. The language processing system according to claim 9, further comprising:

a forbidden word addition section that additionally stores the mis-pronounced word in the forbidden word memory part as the forbidden word pronunciation.