Automated transformation from American English to British English

• US 20020173966A1
• Filed: 12/23/2000
• Published: 11/21/2002
• Est. Priority Date: 12/23/2000
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Tellme Networks Incorporated ([24]7.ai, Inc.)
• Inventors: Henton, Caroline G.
• Granted Patent: US 6,738,738 B2
• Time in Patent Office: Days
• US Class Current: 704/277
• CPC Class Codes: G06F 40/232 Orthographic correction, e.g. spell checking or vowelisation
• CPC Class Codes: G10L 15/1815 Semantic context, e.g. disambiguation of the recognition hypotheses based on word meaning
• CPC Class Codes: G10L 15/187 Phonemic context, e.g. pronunciation rules, phonotactical constraints or phoneme n-grams

Abstract

A method of transforming a voice application program designed for US English speakers to a voice application program for UK English speakers using a computer system is described. In one embodiment, scripts and grammars associated with the voice application program are converted from US-to-UK English. The process includes spelling normalization, lexical normalization, and pronunciation conversion (including where appropriate accounting for stress shifts). The result is necessary word pronunciations for speech recognition of UK English speaker (especially for proper nouns) as well as a script that has been conformed to use UK English spelling and lexical conventions. Additionally, the script can be annotated with pronunciations as a part of the process. Further, in one embodiment a web based interface to the conversion process is provided either standalone or as part of a voice application development environment.

Claims

1. A method of generating a British English phonemic transcription from a word and an American English phonemic transcription of the word using a computer system:

providing the word to a first computer program, the first computer program transforming the spelling of the word according to British English spelling conventions;

providing the word to a second computer program, the second computer program for lexically transforming the word to a semantically equivalent word used in British English and marking the word when no semantic equivalent is found and when the semantic equivalent cannot be identified; and

applying phoneme conversion rules with a third computer program to transform the American English phonemic transcription to the British English phonemic transcription.

2. The method of claim 1, wherein the first computer program, the second computer program and the third computer program operate without use of a dictionary of British English or a comprehensive word list for British English showing one or more of spelling, pronunciation, and word stress.

3. The method of claim 1, wherein the first computer program uses a table comprising rules applicable to certain orthographic forms of the word to produce the British English Spelling.

4. The method of claim 1, wherein the first computer program includes an ad hoc table of a limited number of ad hoc spelling differences between American and British English that are not easily expressed in meaningful orthographic transformation rules, the table not including more than one hundred (100) words.

5. The method of claim 1, wherein the second computer program includes an ad hoc table of ad hoc lexical differences between American and British English, the ad hoc table including lists of words specific to specific areas selected from the set of financial terms, units of measure, musical notation, automotive terms, betting terms, botanical and zoological terms, food names, slang terms, cricket and sports terms, traffic terms, and/or miscellaneous common terms.

6. The method of claim 1, wherein when the second computer program transforms the word to a semantically equivalent word, the American English phonemic transcription is similarly replaced with an American English phonemic transcription for the semantically equivalent word.

7. The method of claim 1, wherein when the second program provides for interactive input and adjustment to marked words prior to the application of the third computer program.

8. The method of claim 1, wherein the American English phonemic transcription represented according to the Computer Phonetic Alphabet (CPA) for American English and wherein the British English phonemic transcription represented according to the CPA for British English.

9. The method of claim 8, wherein the CPA representations of phonemic transcriptions indicate primary and secondary word stress.

10. The method of claim 9, wherein the CPA representation of primary and secondary word stress uses lower ASCII character symbols that are not used in either the CPA for American English or the CPA for British English.

11. The method of claim 9, wherein the third computer program uses changed word stress between American English and British English to transform the American English phonemic transcription to the British English phonemic transcription.

12. The method of claim 11, wherein changed word stress determined according to one or more rules and an ad hoc table of changed word stress.

14. The apparatus of claim 13, further comprising a fifth means for generating an augmented script, the augmented script including British English pronunciation for at least one word.

15. The apparatus of claim 13, further comprising a fifth means for providing a web based interface to the first, second, third and fourth means and wherein the third means is adapted to perform in a web based environment.

16. The apparatus of claim 13, further comprising a fifth means for automatically applying the first, second, third, and fourth means to a voice application without an explicit request to transform the voice application.

18. The method of claim 17, wherein the at least two lower ASCII characters are not used by at least two different CPA representations

19. The method of claim 17, wherein the at least two different CPA representations comprised American English CPA and British English CPA.

20. The method of claim 17, wherein the at least two lower ASCII characters comprised “

1”

(ASCII 0×

31) and “

2”

(ASCII 0×

32).

21. The method of claim 17, further comprising removing the at least two characters from a CPA phonemic transcription prior to providing the transcription to a computer program that does not support a CPA representation of word stress.

22. The method of claim 17, further comprising replacing the at least two characters from a CPA phonemic transcription with corresponding characters for a computer program that does supports a CPA representation of word stress using different characters.

13. An apparatus for transforming a voice application prepared for American English speakers for use by British English speakers, the apparatus comprising:

a first means for analyzing the voice application to identify one or more scripts and one or more grammars, the one or more scripts corresponding to textual material for presentation by one of a text-to-speech system and a human voice talent, the one or more grammars corresponding to descriptions of words that the voice application must be capable of responding to in one or more states;

a second means for automatically performing spelling and lexical normalization of the one or more scripts and the one or more grammars from American English to British English, the normalization producing one or more predetermined markings, the markings being indicative of words likely to require manual review;

a third means for permitting manual review of the normalized scripts and grammars; and

a fourth means for performing phonemic conversions for words in the scripts and the grammars from American English pronunciations to British English pronunciations

17. A method of extending the CPA to indicate word stress in a phonemic transcription represented in the computer phonetic alphabet (CPA), the method comprising:

selecting at least two lower ASCII characters;

selecting a first of the two characters to indicate primary word stress and a second of the two characters to indicate secondary word stress; and

representing word stress to computer programs in CPA phonemic transcriptions by placing the two characters in front of the CPA symbol for the start of the stressed portion of the word.