摘要
现代田野语音学研究中，需要大量字词的语音做分析，而人工切音是费时费力，因而本文设计并实现了一个快捷、准确的音段自动标记切分系统。该系统采用了基于短时能量和短时过零率的端点检测的思想，结合了音节间的时长特点．对音段自动进行添加起始位置标记，再根据录音文本进行自动命名保存即可实现自动切分。
In modem study of field phonetics, it is time consuming and strenuous to cut a great amount of speeches by manuaJ, so the paper designs and realizes a rapid and accurate system of automatic syllabic segment cutting. Combined with inter-syUable time-period characteristics, the system adopts the concept of endpoint detection which based on short-time energy and short-time zero-pass ratio, then added beginning and ending label in syllabic segment automatically, and finally realized auto-cutting by auto-named restore according to recorded text.
出处
《电脑知识与技术》
2008年第5期737-740,共4页
Computer Knowledge and Technology
基金
国家自然科学基金项目（60773052）
研究生创新项目“特定人的藏语孤立词语音识别研究”（ycx08053）.
关键词
语音
端点检测
标记
切分
phonetic voice
endpoint detection, label, cutting
