Dissertation > Industrial Technology > Radio electronics, telecommunications technology > Communicate > Electro-acoustic technology and speech signal processing > Speech Signal Processing > Speech synthesis

Research and Implementation of Emotional Speech Synthesis System

Author RenPengHui
Tutor ZhangXueYing
School Taiyuan University of Technology
Course Information and Communication Engineering
Keywords emotional apeech synthesis emotional speech database prosodicfeatures Hidden Markov Model
CLC TN912.33
Type Master's thesis
Year 2013
Downloads 78
Quotes 0
Download Dissertation

With the continuous development of science and technology, speech synthesis technology has been widely used in many field of human-computer interaction. Emotional speech synthesis as an important branch of research is how to make your computer like people send out the voice with emotional color. This new research direction cause for world-wide researchers raised widespread concern. So emotional speech synthesis technology have been developing rapidly.In this paper, the work done by the following points:①Firstly, the paper summarizes and analyzes the current domestic and foreign research on emotion theory. Choose sadness, happiness, anger, and surprise four kinds of typical basic emotion from a variety of classification standards as a research object. The emotional database is designed and constructed for speech research requirement,which included four kinds of emotions:sadness, happiness, anger, and surprise, each of1000sentences, so all of them were4000sentences about the emotional speech. Firstly, using greedy algorithm to select corpus; And then induced a variety of emotions by Picture induced method; Finally an improved fuzzy comprehensive evaluation method is applied to the speech database from six sides comprehensive evaluation, which are emotional expression, clarity, fluency, scene sense, naturalness and noise effect. Therefore, it can objectively and accurately prove the reliability of the corpus. The construction of the emotional database for emotion speech computing provides an important application foundation and prerequisite work.②Then analyzed several current common speech synthesis method. Synthesis method based on hidden Markov models which modeling time is short, can be automatic without human intervention, synthesized speech has high advantages of naturalness and fluency. So this paper select the synthesis technology to achieve emotional speech synthesis.③Through the study of actual experimental data, summary analysis of sadness, happiness, anger and surprise under four kinds of emotional pitch, duration and amplitude of prosodic features of laws. According to this law, summed up a set of emotional voice modification method. Finally, this paper design a trainable HMM-based emotional speech synthesis system which Include HMM-based trainable speech synthesis system, the emotional analysis module and parameter modification module. This paper has also been verified by experimental reliability and practicality of the system.

Related Dissertations
More Dissertations