Translating between spoken and written language

• US 9,195,650 B2
• Filed: 09/23/2014
• Issued: 11/24/2015
• Est. Priority Date: 05/13/2011
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Inventors: Basson, Sara H., Hamilton, Rick A. II, Jiang, Dan Ning, Kanevsky, Dimitri, Nahamoo, David, Picheny, Michael A., Ramabhadran, Bhuvana, Sainath, Tara N.
• Primary Examiner(s): VO, HUYEN X
• Application Number:
  US14/494,469
• Publication Number:
  US 20150120275A1
• Time in Patent Office: 427 Days
• Field of Search: 704 1- 10, 704/250, 704/251, 704/255, 704/257, 704/277, 704/270
• US Class Current: 1/1
• CPC Class Codes: G06F 40/30 Semantic analysis
• CPC Class Codes: G06F 40/40 Processing or translation of natural language natural language analysis G06F40/20; semantic analysis G06F40/30
• CPC Class Codes: G06F 40/58 Use of machine translation, e.g. for multi-lingual retrieval, for server-side translation for client devices or for real-time translation
• CPC Class Codes: G10L 15/063 Training
• CPC Class Codes: G10L 15/26 Speech to text systems G10L15/08 takes precedence

Abstract

Techniques for converting spoken speech into written speech are provided. The techniques include transcribing input speech via speech recognition, mapping each spoken utterance from input speech into a corresponding formal utterance, and mapping each formal utterance into a stylistically formatted written utterance.

Claims

1. A method for translating a first text in a first language style into a second text in a second language style, the method comprising acts of:

performing, by at least one processor, natural language understanding (NLU) processing on the first text to obtain at least one semantic representation of the first text; and

using, by the at least one processor, the at least one semantic representation to generate the second text in the second language style, the second language style being different from the first language style, wherein the act of using the at least one semantic representation to generate the second text comprises;

identifying, from the first text, at least one first word or phrase that occurs in a first corpus, wherein the first corpus comprises a first plurality of texts in the first language style;

mapping the at least one first word or phrase to at least one second word or phrase that occurs in a second corpus, wherein the second corpus comprises a second plurality of texts in the second language style, the second plurality of texts in the second corpus being in a same language as the first plurality of texts in the first corpus; and

using the at least one semantic representation and the at least one second word or phrase to generate the second text.

2. The method of claim 1, wherein the at least one semantic representation comprises a semantic tree, and wherein performing the NLU processing comprises using an NLU parser to parse the first text to obtain the semantic tree.

3. The method of claim 2, wherein using the at least one semantic representation to generate the second text comprises using the semantic tree in conjunction with a semantic lexicon to generate the second text.

4. The method of claim 1, wherein the NLU processing comprises:

creating a plurality of candidate semantic representations of the first text; and

selecting the at least one semantic representation from the plurality of candidate semantic representations.

5. The method of claim 1, wherein the first language style comprises a spoken language style and the second language style comprises a written language style.

6. The method of claim 1, wherein the first language style comprises a language style associated with a first cultural background and the second language style comprises a language style associated with a second cultural background.

7. At least one non-transitory computer readable storage medium having encoded thereon instructions that, when executed by at least one processor, cause the at least one processor to perform a method for translating a first text in a first language style into a second text in a second language style, the method comprising acts of:

performing, by the at least one processor, natural language understanding (NLU) processing on the first text to obtain at least one semantic representation of the first text; and

using, by the at least one processor, the at least one semantic representation to generate the second text in the second language style, the second language style being different from the first language style, wherein the act of using the at least one semantic representation to generate the second text comprises;

identifying, from the first text, at least one first word or phrase that occurs in a first corpus, wherein the first corpus comprises a first plurality of texts in the first language style;

mapping the at least one first word or phrase to at least one second word or phrase that occurs in a second corpus, wherein the second corpus comprises a second plurality of texts in the second language style, the second plurality of texts in the second corpus being in a same language as the first plurality of texts in the first corpus; and

using the at least one semantic representation and the at least one second word or phrase to generate the second text.

8. The at least one non-transitory computer readable storage medium of claim 7, wherein the at least one semantic representation comprises a semantic tree, and wherein performing the NLU processing comprises using an NLU parser to parse the first text to obtain the semantic tree.

9. The at least one non-transitory computer readable storage medium of claim 8, wherein using the at least one semantic representation to generate the second text comprises using the semantic tree in conjunction with a semantic lexicon to generate the second text.

10. The at least one non-transitory computer readable storage medium of claim 7, wherein the NLU processing comprises:

creating a plurality of candidate semantic representations of the first text; and

selecting the at least one semantic representation from the plurality of candidate semantic representations.

11. The at least one non-transitory computer readable storage medium of claim 7, wherein the first language style comprises a spoken language style and the second language style comprises a written language style.

12. The at least one non-transitory computer readable storage medium of claim 7, wherein the first language style comprises a language style associated with a first cultural background and the second language style comprises a language style associated with a second cultural background.

13. A system for translating a first text in a first language style into a second text in a second language style, the system comprising:

at least one memory storing executable instructions; and

at least one processor programmed by the executable instructions stored in the at least one memory to;

perform natural language understanding (NLU) processing on the first text to obtain at least one semantic representation of the first text; and

use the at least one semantic representation to generate the second text in the second language style, the second language style being different from the first language style, wherein the at least one processor is programmed to generate the second text at least in part by;

identifying, from the first text, at least one first word or phrase that occurs in a first corpus, wherein the first corpus comprises a first plurality of texts in the first language style;

mapping the at least one first word or phrase to at least one second word or phrase that occurs in a second corpus, wherein the second corpus comprises a second plurality of texts in the second language style, the second plurality of texts in the second corpus being in a same language as the first plurality of texts in the first corpus; and

using the at least one semantic representation and the at least one second word or phrase to generate the second text.

14. The system of claim 13, wherein the at least one semantic representation comprises a semantic tree, and wherein the NLU processing comprises using an NLU parser to parse the first text to obtain the semantic tree.

15. The system of claim 14, wherein the at least one processor is programmed to use the semantic tree in conjunction with a semantic lexicon to generate the second text.

16. The system of claim 13, wherein the NLU processing comprises:

creating a plurality of candidate semantic representations of the first text; and

selecting the at least one semantic representation from the plurality of candidate semantic representations.

17. The system of claim 13, wherein the first language style comprises a spoken language style and the second language style comprises a written language style.

18. The system of claim 13, wherein the first language style comprises a language style associated with a first cultural background and the second language style comprises a language style associated with a second cultural background.