现代维吾尔语情感对话语音数据库的设计研究

Corpus Design for Modern Uyghur Affective Speech

在世界学术领域,语音的情感研究处于摸索阶段,建立研究者认可的高质量情感语音数据库是亟待解决的问题。文章通过设计合格语料,建立现代维吾尔语情感语音数据库。以高兴、生气、恐惧、中性和悲伤为基本情感,收录男女声情感对话的语音文件、声学特征文件和标注维吾尔语韵律等语音标注文件。对该数据库进行主观评价的结果表明:语音情感状态自然,能够满足语音情感识别的需求。数据库提供的不同情感状态下维吾尔语韵律特征与声学特征数据,可以供情感语音识别研究、情感合成研究和说话人识别研究。

Affective speech plays a significant role in the study of experimental phonetics.In the world academic circle,it has left a big issue for researchers to build up a qualified affective speech database.The present study focuses on corpus establishment for modern Uyghur affective speech by collecting decent data with five basic emotion(happiness,anger,fear and sadness).There are datasets of conversational speech emotion in Uyghur,annotation files with prosodic and acoustic features.