摘要
由于电子化语言资源较为匮乏,缅甸语语音合成的研究相对滞后.针对该问题,以开发缅甸语语音合成应用系统为目的,设计并实现了一个基于HMM的语音合成系统.首先,为了使计算机按输入文本合成出正确的读音,提出并设计了缅甸语的注音方案,其中重点解决了缅甸语中的变音和变调问题;然后,根据缅甸语的语音特点选取声母及带声调的韵母作为合成基元,并按此设计上下文属性和问题集;最后,基于HTS平台,完整实现了音子自动切分、HMM声学模型训练及语音合成.实验结果表明,该缅甸语语音合成系统具有可行性,可作为后续研究的基线系统.
Due to the lack of electronic language resources,the study of Burmese speech synthesis is lagging behind.To solve this problem,this paper aims to develop a Burmese speech synthesis application system,designs and implements a HMM-based speech synthesis system.First,in order for the computer to synthesize the correct pronunciation according to the input text,a Burmese grapheme-to-phoneme conversion method is proposed and designed,which focuses on solving the problem of pronunciation change in Burmese.Then,based on the language characteristics of Burmese,initials and finals are used as synthesis unit,and context attributes and problem sets are designed.Finally,based on the HTS platform,automatic phoneme segmentation,HMM acoustic model training,and speech synthesis are implemented.Experimental results show that the Burmese speech synthesis system is feasible and can be used as a baseline system for subsequent research.
作者
刘梦媛
杨鉴
LIU Meng-yuan;YANG Jian(School of Information Science and Engineering,Yunnan University,Kunming 650500,China)
出处
《云南大学学报（自然科学版）》
CAS
CSCD
北大核心
2020年第1期19-27,共9页
Journal of Yunnan University(Natural Sciences Edition)
基金
国家自然科学基金(61961043)
关键词
缅甸语
语音合成
文本注音
音子自动切分
声学模型
Burmese
speech synthesis
grapheme-to-phoneme conversion
automatic phoneme segmentation
acoustic model
