Work

MRI-based 3D Model of Spoken Lips

Year: 2016

Type: conference-paper

Abstract: Modeling accurate speech organs are vital to speech synthesis.This paper introduces a construction method for geometric model of spoken lips based on a MRI database.Volume images of articulators were acquired by MRI on one Chinese subject uttering a corpus of sustained articulations.Then 3D models of each organ were annotated from relevant image slices.Finally linear component analysis was used to extract main control parameters from the 3D models of the spoken lips.The result shows that these control parameters could effectively represent the position and motion of spoken lips with a minor reconstruction error (less than 0.15cm).

Source: International Conference on Electrical & Electronics Engineering and Computer Science (ICEEECS)

DOI: 10.2991/iceeecs-16.2016.130

Authors: Yun Chen, Qiang Fang, Wenhuan Lu, Jianguo Wei

Institutions: Tianjin University, Chinese Academy of Social Sciences, Institute of Linguistics

Language: English

Topic: Speech and Audio Processing

Subfield: Signal Processing

Field: Computer Science

Domain: Physical Sciences

Open Access status: hybrid

Funder: National Natural Science Foundation of China

Awards: G1160910298, 基于自适应频率尺度变换的骨导鼾声识别关键技术研究