Techniques for aiding speech-to-speech translation

• US 20070043567A1
• Filed: 08/22/2005
• Published: 02/22/2007
• Est. Priority Date: 08/22/2005
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Zhou, Bowen, Gao, Yuqing, Kuo, Hong-Kwang
• Granted Patent: US 7,552,053 B2
• Time in Patent Office: Days
• US Class Current: 704/257
• CPC Class Codes: G06F 40/45 Example-based machine translation; Alignment
• CPC Class Codes: G10L 15/26 Speech to text systems G10L15/08 takes precedence

Abstract

Techniques for assisting in translation are provided. A speech recognition hypothesis is obtained, corresponding to a source language utterance. Information retrieval is performed on a supplemental database, based on a situational context, to obtain at least one word string that is related to the source language utterance. The speech recognition hypothesis and the word string are then formatted for display to a user, to facilitate an appropriate selection by the user for translation.

Claims

1. A computer-implemented method for translation of a source language utterance of a first user, comprising the steps of:

obtaining a speech recognition hypothesis corresponding to said source language utterance;

performing information retrieval on a supplemental database, based on a situational context, to obtain at least one word string related to said source language utterance; and

formatting said speech recognition hypothesis and said word string for display to facilitate an appropriate translation selection.

2. The method of claim 1, further comprising the additional steps of:

displaying said speech recognition hypothesis and said word string; and

obtaining said translation selection, from said first user, among said speech recognition hypothesis and said word string.

3. The method of claim 2, further comprising the additional step of performing machine translation on said speech recognition hypothesis to obtain a translation result.

4. The method of claim 3, further comprising the additional step of performing text-to-speech synthesis on said translation result.

5. The method of claim 2, wherein said word string has a stored translation, further comprising the additional step of, responsive to said first user'"'"'s translation selection of said word string, performing text-to-speech synthesis on said stored translation.

6. The method of claim 2, further comprising the additional steps of:

obtaining a supplemental first user selection of at least one word in at least one of said speech recognition hypothesis and said word string; and

responsive to said supplemental first user selection, generating at least one candidate choice for a subsequent word string to be translated.

7. The method of claim 2, further comprising the additional steps of:

predicting at least one future time word string likely to be useful in a future time, based on a dialog model of past dialog history and sentence correlation; and

displaying said at least one future time word string in a format to facilitate selection, by said first user, of said future time word string for translation.

8. The method of claim 7, wherein said at least one future time word string is displayed so as to guide said first user to employ a desirable lexicon in subsequent dialog.

9. The method of claim 2, wherein said method is for interaction with said first user in said source language and a second user in a second language, further comprising the additional steps of:

capturing cues from said first user and said second user; and

based on said cues, predicting at least one next word string likely to be desired to be translated next by at least one of said first and second users.

10. The method of claim 9, further comprising the additional step of producing a summary log, of interactions of said first and second users, in said source language and said second language.

11. The method of claim 2, wherein:

said step of performing information retrieval further comprises the step of obtaining multiple word strings related to said source language utterance; and

said displaying step further comprises displaying said speech recognition hypothesis and said word strings together with associated confidence scores.

12. The method of claim 11, further comprising the additional step of performing machine translation on said speech recognition hypothesis to obtain a translation result, wherein:

said word strings have word strings translations associated therewith; and

said displaying step further comprises displaying said speech recognition hypothesis and said word strings together with, respectively, said translation result and associated ones of said word string translations.

13. The method of claim 1, wherein said supplemental database comprises a parallel corpus of indexed training data, including previously translated source language word strings having corresponding word string translations, previously translated into a target language, associated therewith.

14. The method of claim 13, further comprising the additional step of performing automatic speech recognition on said source language utterance to obtain said speech recognition hypothesis.

15. The method of claim 14, wherein:

said automatic speech recognition is performed to obtain a speech N-best result; and

said step of performing information retrieval comprises;

taking said speech N-best result as a query in a weighted combination; and

searching said indexed training data based on said query.

16. The method of claim 1, wherein said at least one word string is closely matched to said source language utterance.

17. The method of claim 1, wherein said situational context includes said speech recognition hypothesis.

18. A computer-implemented method for assisting machine translation, comprising the steps of:

monitoring a dialog of at least one user with a translation apparatus;

analyzing said dialog to develop at least one suggestion for a next word string desired to be translated; and

formatting said next word string for display to said at least one user.

19. A computer program product comprising a computer useable medium having computer useable program code for translation of a source language utterance of a first user, said computer program product including:

computer useable program code for obtaining a speech recognition hypothesis corresponding to said source language utterance;

computer useable program code for performing information retrieval on a supplemental database, based on a situational context, to obtain at least one word string related to said source language utterance; and

computer useable program code for formatting said speech recognition hypothesis and said word string for display to facilitate an appropriate translation selection by said first user.

20. A computer program product comprising a computer useable medium having computer useable program code for assisting machine translation, said computer program product including:

computer useable program code for monitoring a dialog of at least one user with a translation apparatus;

computer useable program code for analyzing said dialog to develop at least one suggestion for a next word string desired to be translated; and

computer useable program code for formatting said next word string for display to said at least one user.