Artificial language generation and evaluation

• US 20020198712A1
• Filed: 06/11/2002
• Published: 12/26/2002
• Est. Priority Date: 06/12/2001
• Status: Abandoned Application

• Current Assignee: Hewlett-Packard Development Company, L.P. (HP Inc.)
• Original Assignee: Hewlett-Packard Company (HP Inc.)
• Inventors: Belrose, Guillaume, Hinde, Stephen John
• Application Number:
  US10/165,973
• Publication Number:
  US 20020198712A1
• Time in Patent Office: Days
• US Class Current: 704/251
• CPC Class Codes: G10L 13/02 Methods for producing synthetic speech; Speech synthesisers
• CPC Class Codes: G10L 15/06 Creation of reference templates; Training of speech recognition systems, e.g. adaptation to the characteristics of the speaker's voice G10L15/14 takes precedence

Abstract

A method is provided of generating an artificial language for use, for example, in human speech interfaces to devices. In a preferred implementation, the language generation method involves using a genetic algorithm to evolve a population of individuals over a plurality of generations, the individuals forming or being used to form candidate artificial-language words. The method is carried in a manner favouring the production of artificial-language words which are more easily correctly recognised by a speech recognition system and have a familiarity to a human user. This is achieved, for example, by selecting words for evolution on the basis of an evaluation carried out using a fitness function that takes account both of correct recognition of candidate words when spoken to a speech recognition system, and the similarity of candidate words to words in a set of user-favourite words.

Claims

1. A method of automatically generating candidate artificial-language words, the method involving a process that is specifically set to favour artificial-language words which are more easily correctly recognised by a speech recognition system and have a familiarity to a human user.

2. A method according to claim 1, wherein said process involves evaluating words both in terms of how easily they are correctly recognised by a speech recognition system and of a familiarity to a human user.

3. A method according to claim 2, wherein the evaluation of words in terms of how easily they are correctly recognised by a speech recognition system is effected by presenting the words to a speech recognition system and measuring the resultant recognition performance.

4. A method according to claim 2, wherein the evaluation of words in terms of how easily they are correctly recognised by a speech recognition system is effected by analysis of the phoneme composition of the words in relation to a confusion matrix established for a target speech recognition system.

5. A method according to claim 2, wherein the evaluation of words in terms of a familiarity to a human user is effected by presenting the words to a speech recognition system set to recognise a set of reference words familiar to a user and measuring the resultant recognition performance.

6. A method according to claim 2, wherein the evaluation of words in terms of a familiarity to a human user is effected by analysis of the phoneme composition of the words in relation to that of a set of reference words familiar to a user.

7. A method according to claim 1, wherein said process involves creating words in a manner favouring words that are more easily recognised by a speech recognition system and evaluating the words thus created in terms of a familiarity to a human user.

8. A method according to claim 7, wherein the evaluation of words in terms of a familiarity to a human user is effected by presenting the words to a speech recognition system set to recognise a set of reference words familiar to a user and measuring the resultant recognition performance.

9. A method according to claim 7, wherein the evaluation of words in terms of a familiarity to a human user is effected by analysis of the phoneme composition of the words in relation to that of a set of reference words familiar to a user.

10. A method according to claim 7, wherein the creation of words in a manner favouring words that are more easily recognised by a speech recognition system, is effected by choosing phoneme and phoneme combinations which according to a confusion matrix established for a target speech recognition system, are less likely to be confused.

11. A method according to claim 1, wherein said process involves creating words in a manner favouring words that have a familiarity to a human user, and evaluating the words thus created in terms of how easily they are correctly recognised by a speech recognition system.

12. A method according to claim 11, wherein the evaluation of words in terms of how easily they are correctly recognised by a speech recognition system is effected by presenting the words to a speech recognition system and measuring the resultant recognition performance.

13. A method according to claim 11, wherein the evaluation of words in terms of how easily they are correctly recognised by a speech recognition system is effected by analysis of the phoneme composition of the words in relation to a confusion matrix established for a target speech recognition system.

14. A method according to claim 11, wherein the creation of words in a manner favouring words that have a familiarity to a user, is effected by using phonemes and/or phoneme combinations from a set of reference words familiar to a user, or like-sounding phonemes and/or phoneme combinations.

15. A method according to claim 1, wherein said process involves creating words in a manner favouring words that are more easily recognised by a speech recognition system favouring and have a familiarity to a human user.

16. A method according to claim 15, wherein the creation of words in a manner favouring words that are more easily recognised by a speech recognition system, is effected by choosing phoneme and phoneme combinations which according to a confusion matrix established for a target speech recognition system, are less likely to be confused.

17. A method according to claim 15, wherein the creation of words in a manner favouring words that have a familiarity to a user, is effected by using phonemes and/or phoneme combinations from a set of reference words familiar to a user, or like-sounding phonemes and/or phoneme combinations.

18. A method according to claim 1 wherein said familiarity is that of sounding similar to a natural language word.

19. A method according to claim 1, wherein at least selected ones of the generated artificial language words are stored on a transferable storage medium.

20. A method of conditioning a speech recogniser, comprising the steps of:

generating words of an artificial language using a method according to claim 1, and

loading the generated artificial-language words into a lexicon of the speech recogniser.

21. A method of conditioning a speech recogniser, comprising the steps of:

generating words of an artificial language using a method according to claim 1, and

training the speech recogniser to recognise the generated artificial-language words.

22. A transferable storage medium to which a set of artificial-language words have been stored in accordance with claim 19.

23. A speech recogniser conditioned to recognise artificial-language words according to the method of claim 20.

24. A speech recogniser conditioned to recognise artificial-language words according to the method of claim 21.

25. A set of artificial-language words created by the method of claim 1.

27. A method according to claim 26, wherein the artificial-language words are spoken to the speech recognition system by multiple text-to-speech converters in turn, the fitness measures made in respect of any particular word being a combination of the measures made for the speaking of the word by each converter.

28. A method according to claim 26, wherein the artificial-language words are spoken by a text-to-speech conversion system to the speech recogniser system, the channel involving these systems being implemented in a manner such that said fitness measure takes account of at least one desired operational characteristic.

29. A method according to claim 28, wherein said at least one desired operational characteristic is at least one of:

gender independence, for which purpose the text-to-speech system is provided with multiple text-to-speech converters corresponding to different genders to generate spoken versions of the words;

acoustic independence, for which purpose the speech recognizer system is provided with multiple speech recognizers corresponding to different acoustic models;

robustness to noise, for which purpose noise is introduced into the channel.

26. A method of evaluating words of an artificial language in respect of their usage as a spoken human language for a man-machine interface, the method involving applying a fitness function to each artificial-language word where said fitness function comprises a combination of:

a measure of the ease of correct recognition of a candidate artificial-language word when spoken to a speech recognition system; and

a measure of the similarity of a candidate artificial-language word to any constituent word of a set of reference words as measured by a speech recognition system to which said word is spoken.