| Paper: | SP-P16.2 | ||
| Session: | Speech Modeling for Robust Speech Recognition | ||
| Time: | Friday, May 21, 15:30 - 17:30 | ||
| Presentation: | Poster | ||
| Topic: | Speech Processing: Robust Speech Recognition | ||
| Title: | TONE ARTICULATION MODELING FOR MANDARIN SPONTANEOUS SPEECH RECOGNITION | ||
| Authors: | Jian-Lai Zhou; Microsoft Research Asia | ||
| Ye Tian; Microsoft Research Asia | |||
| Yu Shi; Microsoft Research Asia | |||
| Chao Huang; Microsoft Research Asia | |||
| Eric Chang; Microsoft Research Asia | |||
| Abstract: | Tone modeling is an unavoidable problem in Mandarin speech recognition. It is accepted that the tone is determined by pitch contour. In continuous speech, the pitch contour exhibits variable patterns, and it is influenced by its tone context dramatically. Although several effective methods have been proposed to improve the accuracy of tonal syllable in Mandarin continue speech recognition, many recognition errors resulted from poor discrimination of the tone of the acoustic model [1][2][3][4]. Furthermore, the case in casual speech recognition becomes worse. In this paper, we will report our work on tone articulation modeling. Tone context dependent model is used to model unstable pitch pattern caused by co-articulation in continuous speech. Corresponding acoustic features were investigated as well. Our methods were evaluated on two test sets: one is reading style speech data, another is spontaneous. The experimental results showed that the more casual the test data are, the more effective the proposed method turns out to be. It must be pointed out that the proposed algorithm has not reached its optimal level. Several factors which have potential to improve the proposed method are discussed in the final part in this paper. | ||
| Back | |||
Home -||-
Organizing Committee -||-
Technical Committee -||-
Technical Program -||-
Plenaries
Paper Submission -||-
Special Sessions -||-
ITT -||-
Paper Review -||-
Exhibits -||-
Tutorials
Information -||-
Registration -||-
Travel Insurance -||-
Housing -||-
Workshops