基于韵律特征参数的情感语音合成算法研究  

Synthesis of emotional speech based on prosody parameters

作　　者：何凌[1] 黄华[1] 刘肖珩[2] 

机构地区：[1]四川大学电气信息学院,四川成都610065 [2]四川大学基础医学与法医学院,四川成都610041

出　　处：《计算机工程与设计》2013年第7期2566-2569,2584,共5页Computer Engineering and Design

基　　金：国家自然科学基金项目(10972148)

摘　　要：为了合成更为自然的情感语音,提出了基于语音信号声学韵律参数及时域基音同步叠加算法的情感语音合成系统。实验通过对情感语音数据库中生气、无聊、高兴和悲伤4种情感的韵律参数分析,建立4种情感模板,采用波形拼接语音合成技术,运用时域基音同步叠加算法合成含有目标感情色彩的语音信号。实验结果表明,运用波形拼接算法,调节自然状态下语音信号的韵律特征参数,可合成较理想的情感语音。合成的目标情感语音具有明显的感情色彩,其主观情感类别判别正确率较高。In order to synthesize more natural emotional speech signals, an emotional speech synthesis system was proposed based on the acoustic prosody parameters and time domain pitch synchronous overlap add algorithm. The experiment built up prosodic templates for four emotions： angry, bored, happy and sad, through analyzing the prosody parameters of emotional speech signals. Then the waveform concatenative technique time domain pitch synchronous overlap add algorithm was used to im- plement emotional speech synthesis. The experiment results show that the proposed waveform concatenative algorithm combined with acoustic prosodic parameters modification method, had good performance on emotional speech synthesis. The synthesized emotional speech show strong emotional arouse and high subjective classification accuracy is achieved.

关 键 词：情感语音合成 韵律参数 时域基音同步叠加 波形拼接 基音频率 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]