Multilingual dialogue system and controlling method thereof

• US 20100174523A1
• Filed: 11/30/2009
• Published: 07/08/2010
• Est. Priority Date: 01/06/2009
• Status: Active Grant

• Current Assignee: Samsung Electronics Co. Ltd.
• Original Assignee: Samsung Electronics Co. Ltd.
• Inventors: PARK, Ki Cheol, HAN, Woo Sup, HA, Tae Sin, JANG, Jun Won
• Granted Patent: US 8,484,011 B2
• Time in Patent Office: Days
• US Class Current: 704/2
• CPC Class Codes: G10L 15/18 using natural language modelling

Abstract

A multilingual dialogue system enabling a user to conduct dialogue in various languages, and a controlling method thereof. According to the multilingual dialogue system and the controlling method, reliability of the user'"'"'s language can be evaluated, thereby coping with errors generated during dialogue.

Claims

1. A multilingual dialogue system comprising:

a sound input unit to be input with sound of a language spoken by a user;

a common dialogue module to produce dialogue contents to be supplied to the user in one common language; and

a sound output unit to supply the user with the dialogue contents in the form of sound.

2. The multilingual dialogue system according to claim 1, further comprising a reliability evaluator to evaluate reliability of the dialogue contents.

3. The multilingual dialogue system according to claim 2, wherein the reliability evaluator includes an evaluation function generating unit to generate a numerical value indicating the reliability using a plurality of evaluation factors.

4. The multilingual dialogue system according to claim 3, wherein the evaluation function generating unit generates a reliability evaluation function corresponding to the reliability according to:

f(t)=(AMEV+LMEV1+LMEV2+CAEV)/4

wherein, AMEV refers to a value evaluating the user'"'"'s language using a sound model, being set in a range between 0 and 1, LMEV1 refers to a value evaluating the user'"'"'s language using a language model, being set in a range between 0 and 1, LMEV2 refers to a value evaluating the common language using an other language model, being set in a range between 0 and 1, and CAEV refers to a value evaluating the common language using a situation recognition model, being set in a range between 0 and 1.

5. The multilingual dialogue system according to claim 1, further comprising a key input unit for the user to designate the language spoken by the user.

6. The multilingual dialogue system according to claim 1, wherein the common dialogue module includes a language discerning unit to discern the language being spoken by the user.

7. The multilingual dialogue system according to claim 6, wherein the language discernment unit applies a plurality of language models.

8. The multilingual dialogue system according to claim 7, wherein the plurality of language models include a parallel phone recognition language model (PPRLM) and a Gaussian mixture model (GMM).

9. The multilingual dialogue system according to claim 1, wherein the common dialogue module comprises a translator that translates texts in the user'"'"'s language to texts in the common language and vice versa.

10. The multilingual dialogue system according to claim 2, wherein the common dialogue module selects dialogue contents based on the evaluated reliability.

11. A controlling method for a multilingual dialogue system, comprising:

inputting sound of a user'"'"'s language;

converting the input sound to texts using one common language and producing dialogue contents based on the texts in the common language; and

supplying the user with the dialogue contents in the form of sound.

12. The controlling method according to claim 11, wherein the user directly designates the language spoken by the user.

13. The controlling method according to claim 11, wherein a plurality of language models are used to discern the language being spoken by the user.

14. The controlling method according to claim 13, wherein the plurality of language models include a PPRLM and a GMM.

15. The controlling method according to claim 11, wherein reliability of the dialogue contents is numerically evaluated.

16. The controlling method according to claim 15, wherein the reliability is evaluated using a sound model with respect to the user'"'"'s language, a language model with respect to the user'"'"'s language, an other language model with respect to the common model, and a situation recognition model with respect to the common language.

17. The controlling method according to claim 15, wherein the dialogue contents are selected based on the evaluated reliability.