Voice processing methods and systems

• US 8,543,400 B2
• Filed: 06/06/2008
• Issued: 09/24/2013
• Est. Priority Date: 06/11/2007
• Status: Active Grant

• Current Assignee: National Taiwan University (Government of The Republic of China)
• Original Assignee: National Taiwan University (Government of The Republic of China)
• Inventors: Lee, Lin-Shan, Chang, Chia-Lin, Lin, Yi-Jing, Lin, Che-Kuang, Wang, Yow-Bang, Lee, Yun-Huan, Cheng, Li-Wei
• Primary Examiner(s): Desir, Pierre-Louis
• Application Number:
  US12/135,026
• Publication Number:
  US 20080306738A1
• Time in Patent Office: 1,936 Days
• Field of Search: 704/236, 704/251, 704/254, 704/255, 704/260, 704/270, 704/243
• US Class Current: 704/243
• CPC Class Codes: G09B 19/04 Speaking with audible presentation of the material to be studied G09B5/04
• CPC Class Codes: G09B 19/06 Foreign languages with audible presentation of material to be studied G09B5/04
• CPC Class Codes: G10L 15/26 Speech to text systems G10L15/08 takes precedence

Abstract

Voice processing methods and systems are provided. An utterance is received. The utterance is compared with teaching materials according to at least one matching algorithm to obtain a plurality of matching values corresponding to a plurality of voice units of the utterance. Respective voice units are scored in at least one first scoring item according to the matching values and a personified voice scoring algorithm. The personified voice scoring algorithm is generated according to training utterances corresponding to at least one training sentence in a phonetic-balanced sentence set of a plurality of learners and at least one real teacher, and scores corresponding to the respective voice units of the training utterances of the learners in the first scoring item provided by the real teacher.

Claims

1. A voice processing method, comprising:

receiving an utterance;

comparing the utterance with a corresponding utterance in teaching materials according to at least one matching algorithm to obtain a plurality of matching values corresponding to a plurality of voice units of the utterance;

scoring the respective voice units in at least one first scoring item according to the matching values corresponding to the respective voice units;

scoring the utterance in at least one second scoring item according to the scores of the respective voice units in the first scoring item; and

scoring the utterance in the second scoring item according to the scores of the respective voice units in the first scoring item and at least one personified voice scoring algorithm for the second scoring item, wherein the method for generating the personified voice scoring algorithm for the second scoring item comprises the steps of;

receiving training utterances corresponding to at least one training sentence in a phonetic-balanced sentence set produced by a plurality of learners and at least one real teacher;

comparing the training utterances of the learners and that of the real teacher according to at least one matching algorithm to obtain a plurality of matching values corresponding to a plurality of voice units of the training utterances;

receiving scores corresponding to the respective voice units of the training utterances of the learners in the first scoring item provided by the real teacher;

receiving scores corresponding to the training utterances of the learners in the second scoring item provided by the real teacher; and

determining the personified voice scoring algorithm for the second scoring item according to the scores corresponding to the respective voice units of the training utterances in the first scoring item and the scores corresponding to the training utterances in the second scoring item.

2. The method of claim 1, further comprising aligning the respective voice units of the utterance to that of the teaching materials before the utterance is compared with the corresponding utterance in the teaching materials.

3. The method of claim 1, further comprising determining an overall score of the utterance according to the score of the utterance in the second scoring item.

4. The method of claim 1, wherein the first scoring item comprises the phoneme correctness, the pitch correctness, or the phoneme-length correctness for the respective voice units.

5. The method of claim 1, wherein the second scoring item comprises the phoneme correctness, the pitch correctness, or the phone-speed correctness for the voice.

6. The method of claim 1, further comprising determining whether at least one of the voice units conforms to at least one pronunciation error pattern.

7. The method of claim 6, further comprising determining a correction action according to the pronunciation error pattern when a specific voice unit within the voice units conforms to the pronunciation error pattern.

8. The method of claim 1, further comprising scoring the respective voice units in the first scoring item according to the matching values and at least one personified voice scoring algorithm, wherein the method for generating the personified voice scoring algorithm comprises the steps of:

receiving training utterances corresponding to at least one training sentence in a phonetic-balanced sentence set produced by a plurality of learners and at least one real teacher;

comparing the training utterances of the learners and that of the real teacher according to at least one matching algorithm to obtain a plurality of matching values corresponding to a plurality of voice units of the training utterances;

receiving scores corresponding to the respective voice units of the training utterances of the learners in the first scoring item provided by the real teacher; and

determining the personified voice scoring algorithm for the first scoring item according to the matching values and the scores corresponding to the respective voice units of the training utterances in the first scoring item.

9. The method of claim 8, wherein the method for generating the phonetic-balanced sentence set comprises the steps of:

setting a plurality of phonetic characteristics for a target language; and

searching for the at least one training sentence that covers the phonetic characteristics from the teaching materials to form the phonetic-balanced sentence set.

10. The method of claim 9, wherein the phonetic characteristics comprises all voice units or pronunciation variations for the target language.

11. The method of claim 1, further comprising determining an overall score of the voice according to the score of the voice in the second scoring item and a personified voice scoring algorithm for the overall score, wherein the personified voice scoring algorithm for the overall score is determined according to the scores corresponding to the training utterances of the learners in the second scoring item provided by the real teacher.