Kiosk for multiple spoken languages

• US 6,061,646 A
• Filed: 12/18/1997
• Issued: 05/09/2000
• Est. Priority Date: 12/18/1997
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Paulsen, Robert Charles Jr., Martino, Michael John
• Primary Examiner(s): Isen, Forester W.
• Assistant Examiner(s): EDOUARD, PATRICK NESTOR
• Application Number:
  US08/993,606
• Time in Patent Office: 873 Days
• Field of Search: 704/1, 704/2-7, 704/254, 704/257, 704/270, 704/275, 704/277, 379/88.04, 379/88.05, 379/88.06
• US Class Current: 704/3
• CPC Class Codes: G06F 40/263 Language identification

Abstract

The method for providing information in response to a question in one of a plurality of natural spoken languages begins by recognizing a detected utterance with a speech recognition engine equipped with a plurality of small dictionaries. Each of the small dictionaries is for respective one of the plurality of languages. Each small dictionary including speech data for a selected few common words in the respective language. Next, the method selects one of the plurality of languages as the language of the detected utterance based on a number of recognized words for each language from the small dictionaries. Next, a more thorough recognition of the detected utterance using a large dictionary for the language of the detected utterance which contains information on a much larger vocabulary. Finally, the method responds to the user in the selected language, i.e. the language of the detected utterance, either aurally or visually. Once the language of a first utterance is identified, a timer is started. Responsive to detecting a new utterance within a predetermined period, the method continues using the large dictionary to recognize the new detected utterance and responding to the user in the language of the detected utterance. If the timer times out, the method reinitializes and a new utterance is tested by all of the small dictionaries.

Claims

1. A method for providing information in response to a question in one of a plurality of natural spoken languages, comprising the steps of:

recognizing a detected utterance with a speech recognition engine equipped with a plurality of small dictionaries each for respective one of the plurality of languages, each small dictionary including speech data for a selected few common words in the respective language;

selecting one of the plurality of languages as the language of the detected utterance based on a number of recognized words for each language from the small dictionaries;

recognizing the detected utterance using a large dictionary for the language of the detected utterance; and

responding to the user in the selected language.

2. The method as recited in claim 1 wherein the response to the user is an aural response in the language of the detected utterance.

3. The method as recited in claim 1 further comprising the steps of:

starting a timer;

responsive to detecting a new utterance within a predetermined period, using the large dictionary to recognize the new detected utterance; and

responding to the user in the language of the detected utterances.

4. The method of claim 3 further comprising the step of responsive to failure to detect a new utterance within the predetermined time period, reinitializing the system so that future utterances are recognized with the plurality of small dictionaries.

5. The method of claim 1 further comprising the step of retrieving the large dictionary from computer storage.

6. The method of claim 1 further comprising the step of storing one large dictionary as a default dictionary in a condition where it can be more quickly accessed than a plurality of other large dictionaries for respective languages.

7. The method as recited in claim 1 wherein the response is a visual display at least in part in the language of the detected utterance.

8. The method as recited in claim 1 wherein a percentage coverage in the respective natural language for each of the small dictionaries of the common words is substantially equivalent to the other small dictionaries.

9. The method as recited in claim 1 wherein the small dictionaries contain most common words in each natural language spoken during a specialized task.

10. The method as recited in claim 1 wherein the large dictionary contains words in the natural language of the detected utterance which are typically spoken during a specialized task.

11. The method as recited in claim 5 further comprising the step of retrieving a specialized speech recognition engine for the natural language of the detected utterance.

12. A system including processor and memory for providing information in response to a question in one of a plurality of natural spoken languages, comprising:

a microphone for detecting an utterance from a user;

a plurality of small dictionaries each for respective one of the plurality of languages, each small dictionary including speech data for a selected few common words in the respective language;

at least one speech recognition engine for recognizing the detected utterance, the at least one speech recognition engine using the plurality of small dictionaries to recognize words within the detected utterance;

a language recognizer for selecting one of the plurality of languages as the language of the detected utterance based on a number of recognized words for each language from the small dictionaries;

a plurality of large dictionaries for the plurality of languages usable by the at least one speech recognition engine for recognizing words within the detected utterance;

retrieval means for retrieving the large dictionary for the selected language; and

means for responding to the user in the selected language.

13. The system as recited in claim 12 further comprising a speaker for responding to the user in an aural response in the language of the detected utterance.

14. The system as recited in claim 12 further comprising a speech recognition engine for each of the plurality of languages.

15. The system as recited in claim 12 further comprising a speech recognition engine for each of a plurality of language classes.

16. The system as recited in claim 12 further comprising a display for visually presenting a response at least a part of which is in the language of the detected utterance.

17. The system as recited in claim 12 wherein a percentage coverage in the respective natural language for each of the small dictionaries of the common words is substantially equivalent to the other small dictionaries.

18. The system as recited in claim 12 wherein the small dictionaries contain most common words in each natural language spoken during a specialized task.

19. The system as recited in claim 12 further comprising a specialized speech recognition engine for utilizing the large dictionary of a particular language for recognizing words of the natural language of the detected utterance.

20. A computer program product in a computer readable medium for providing information in response to a question in one of a plurality of natural spoken languages, comprising:

means for recognizing a detected utterance with according to each of a plurality of small dictionaries each for respective one of the plurality of languages, each small dictionary including speech data for a selected few common words in the respective language;

means for selecting one of the plurality of languages as the language of the detected utterance based on a number of recognized words for each language from the small dictionaries;

means for recognizing the detected utterance using a large dictionary for the language of the detected utterance; and

means for responding to the user in the selected language.

21. The product as recited in claim 21 further comprising a speech synthesis module for responding to the user in the language of the detected utterance.

22. The product as recited in claim 21 further comprising:

a timer means; and

means responsive to detecting a new utterance within a predetermined period for using the large dictionary to recognize the new detected utterance.