面向说话人识别的甘肃临夏少数民族方言语音库的建设

Construction of the Speech Database of Minority Dialect in Linxia, Gansu for Speaker Recognition

面向方言的说话人识别研究刚刚起步,因此建设一个用于说话人识别的、符合地域特色的方言语音库迫在眉睫.本文结合甘肃临夏回族自治州方言的特点,构建一个面向说话人识别的方言语音库.语音库包含普通话和方言音频资料,使用电脑和手机两种方式采集,所有语料内容涉及文化、教育、生活、情感、历史、新闻、评论等,录音者由来自临夏回族自治州的100人组成,共产生了语音语料2400条,为基于甘肃少数民族方言的说话人识别研究奠定了基础.

The study of speaker recognition for dialects has just started,so it is urgent to construct a dialect speech database which is suitable for speaker recognition.Based on the characteristics of the dialect in Linxia Hui Autonomous Prefecture,Gansu Province,this paper constructs a dialect speech database,which contains Mandarin and dialect audio materials,which are collected by computer and mobile phone.All corpus content involves culture,education,life,emotion,history,news,comments,etc.The participants in speech recording is composed of 100 people from Linxia Hui Autonomous Prefecture,and a total of 2400 phonetic corpora have been produced,which lays a foundation for the study of speaker recognition based on Gansu minority dialects.