Text-to-speech

文语转换

A term used in phonetics and computational linguistics to refer to a system of speech synthesis which can transform a conventional orthographic representation of language into its spoken equivalent. Such a system begins by carrying out a morphological and phonological analysis of an input text, taking into account such matters as regular v. irregular forms. Letter-to-sound rules and other special features generate word-level phonological representations, which are then transformed into a phonetic representation (incorporating features of connected speech, including sentence prosody). Synthesis takes place using a rule-based system, the output being provided by a terminal analog synthesizer.

语音学和计算语言学术语，指一个言语合成系统能将语言约定俗成的文字表征转换成对应的语音表征。这样一个系统的运作过程是，先对一输入文本作形态和音系分析，分析时考虑到规则和不规则形式这类问题。然后字母—语音转换规则和其他一些专门特征生成词平面上的音系表征式，后者再转换成语音表征式（容纳连续言语的特征，包括句子韵律）。合成的实现靠一个基于规则的系统，输出由一个终端模拟合成器提供。