Method for increasing dialect precision and usability in speech recognition and text-to-speech systems

• US 6,343,270 B1
• Filed: 12/09/1998
• Issued: 01/29/2002
• Est. Priority Date: 12/09/1998
• Status: Expired due to Term

• Current Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Original Assignee: International Business Machines Corporation
• Inventors: Bahl, Lalit R., Cohen, Paul S.
• Primary Examiner(s): SMITS, TALIVALDIS IVARS
• Application Number:
  US09/207,966
• Time in Patent Office: 1,147 Days
• Field of Search: 704/235, 704/257, 704/260
• US Class Current: 704/257
• CPC Class Codes: G10L 13/08 Text analysis or generation of parameters for speech synthesis out of text, e.g. grapheme to phoneme translation, prosody generation or stress or intonation determination
• CPC Class Codes: G10L 15/065 Adaptation
• CPC Class Codes: G10L 15/187 Phonemic context, e.g. pronunciation rules, phonotactical constraints or phoneme n-grams
• CPC Class Codes: G10L 2015/225 Feedback of the input speech

Abstract

In accordance with the present invention, a method for increasing both dialect precision and usability in speech recognition and text-to-speech systems is described. The invention generates non-linear (i.e. encoded)baseform representations for words and phrases from a pronunciation lexicon. The baseform representations are encoded to incorporate both pronunciation variations and dialectal variations. The encoded baseform representations may be later expanded (i.e. decoded) into one or more linear dialect specific baseform representations, utilizing a set of dialect specific phonological rules.

The method comprises the steps of: constructing an encoded pronunciation lexicon having a plurality of encoded and unencoded baseforms; inputting one or more user specified dialects; selecting dialect specific phonological rules from a rule set database; and decoding the encoded pronunciation lexicon using the dialect specific phonological rules to yield a dialect specific decoded pronunciation lexicon.

Claims

1. A method for generating a dialect specific pronunciation lexicon, the method comprising the steps of:

a) constructing an encoded pronunciation lexicon, said encoded pronunciation lexicon including a plurality of nonlinear baseforms encoded nonlinearly to include one of dialectal and pronunciation alternatives;

b) inputting one or more user specified dialects;

c) selecting phonological rule sets from a rule set database responsive to said one or more user specified dialects; and

d) decoding the encoded pronunciation lexicon by applying the phonological rule sets to the encoded pronunciation lexicon yielding a dialect specific decoded pronunciation lexicon including a plurality of linear dialect specific baseforms.

2. The method of claim 1, wherein the step of constructing an encoded pronunciation lexicon further comprises encoding one or more phones of a plurality of linear baseforms from a pronunciation lexicon.

3. The method of claim 2, wherein the step of encoding at least one phone of the plurality of linear baseforms further comprises encoding said one or more phones with a set of two or more pronunciation alternatives, wherein each alternative is separated by a pronunciation separation symbol.

4. The method of claim 3, wherein the encoding of said one or more phones further includes a start group symbol and an end group symbol for setting off each set of pronunciation alternatives.

5. The method of claim 2, wherein the step of encoding at least one phone of the plurality of linear baseforms further comprises encoding said one or more phones with a set of two or more dialectal alternatives, wherein each alternative is separated by a dialectal separation symbol.

6. The method of claim 5, wherein the encoding of said one or more phones further includes a start group symbol and an end group symbol for setting off each set of dialectal alternatives.

7. The method of claim 1, wherein the rule set database is comprised of a plurality of dialect specific phonological rule sets, wherein each rule set further comprises a plurality of dialect specific phonological rules.

8. The method of claim 1, wherein the phonological rule sets are comprised of a plurality of dialect specific phonological rules, wherein a subset of said phonological rules define how one of said pronunciation alternatives is selected from the set of pronunciation alternatives.

9. The method of claim 1, wherein the phonological rule sets are comprised of a plurality of dialect specific phonological rules, wherein a subset of said phonological rules define how one of said dialectal alternatives is selected from the set of dialectal alternatives.

10. The method of claim 1, wherein the step of inputting one or more user specified dialects further comprises the step of characterizing a speaker'"'"'s accent through the use of diagnostic phrases.

11. The method of claim 1, wherein the step of inputting one or more user specified dialects further comprises the step of interacting with a user via a touch screen display having maps of residence history.

12. The method of claim 1, wherein the step of decoding the pronunciation lexicon using the phonological rules further comprises:

retrieving rule sets from said rule set database corresponding to said user specified dialects; and

applying each of said rules contained within said retrieved rule sets to each of said plurality of encoded baseforms to yield a plurality of decoded baseforms.

13. A dialect specific pronunciation lexicon generating apparatus comprising:

means for constructing an encoded pronunciation lexicon, said pronunciation lexicon including a plurality of baseforms encoded nonlinearly to include one of dialectal and pronunciation alternatives.;

means for inputting one or more user specified dialect preferences; and

means for decoding the encoded pronunciation lexicon.

14. The dialect specific pronunciation lexicon generating apparatus as claimed in claim 13, wherein:

the construction means encodes one or more phones of a plurality of linear baseforms with dialectal and pronunciation variations.

15. The dialect specific pronunciation lexicon generating apparatus as claimed in claim 13, wherein:

the input means comprises a microphone for characterizing a speaker'"'"'s accent through the use of diagnostic phrases.

16. The dialect specific pronunciation lexicon generating apparatus as claimed in claim 13, wherein:

the input means comprises a touch screen display for displaying maps of a speaker'"'"'s residence history.

17. The dialect specific pronunciation lexicon generating apparatus as claimed in claim 13, wherein the decoding means further comprises:

means for selecting one or more dialect specific phonological rule sets from a rule set database;

means for applying said one or more dialect specific phonological rule sets to said encoded pronunciation lexicon.

18. A computer program device readable by a machine, tangibly embodying a program of instructions executable by the machine to perform method steps for generating a dialect specific pronunciation lexicon, the method comprising the steps of:

a) constructing an encoded pronunciation lexicon, said encoded pronunciation lexicon including a plurality of nonlinear baseforms encoded nonlinearly to include one of dialectal and pronunciation alternatives;

b) inputting one or more user specified dialects;

c) selecting phonological rule sets from a rule set database responsive to said one or more user specified dialects; and

d) decoding the encoded pronunciation lexicon by applying the phonological rule sets to the encoded pronunciation lexicon yielding a dialect specific decoded pronunciation lexicon including a plurality of linear dialect specific baseforms.

19. The method of claim 18, wherein the step of constructing an encoded pronunciation lexicon further comprises encoding one or more phones of a plurality of linear baseforms from a pronunciation lexicon.

20. The method of claim 19, wherein the step of encoding at least one phone of the plurality of linear baseforms further comprises encoding said one or more phones with a set of two or more pronunciation alternatives, wherein each alternative is separated by a pronunciation separation symbol.

21. The method of claim 20, wherein the encoding of said one or more phones further includes a start group symbol and an end group symbol for setting off each set of pronunciation alternatives.

22. The method of claim 19, wherein the step of encoding at least one phone of the plurality of linear baseforms further comprises encoding said one or more phones with a set of two or more dialectal alternatives, wherein each alternative is separated by a dialectal separation symbol.

23. The method of claim 18, wherein the rule set database is comprised of a plurality of dialect specific phonological rule sets, wherein each rule set further comprises a plurality of dialect specific phonological rules.

24. The method of claim 18, wherein the step of decoding the pronunciation lexicon using the phonological rules further comprises:

retrieving rule sets from said rule set database corresponding to said user specified dialects; and

applying each of said rules contained within said retrieved rule sets to each of said plurality of encoded baseforms to yield a plurality of decoded baseforms.