Abstract: This paper describes a general-purpose Chinese dialect speech database, which can be applied to speaker information analysis, character-words recognition, speech recognition etc. The speech database, which includes seven kinds of most common Chinese dialects, has reached one hundred and six hours by multi-channel record modes and has already preprocessed. Based on the work, the design criteria and implementation scheme of Chinese dialects speech database are proposed, which is useful for the establishment of Chinese speech database, especially Chinese dialect speech database.

Key words: Chinese dialect speech database, speaker information analysis, character-words recognition

摘要： 建立了一个多用途汉语方言语音数据库，用于说话人信息处理、方言特征词识别、语音识别等领域的研究。以多通道的方式采集时长106小时的语音数据，包括七种主要的汉语方言区语音，对数据进行预处理。在此基础上提出了汉语方言数据库的设计标准以及实施方案，有助于推动汉语语音库特别是方言语音库的建立。

关键词: 汉语方言数据库, 说话人信息处理, 方言特征词识别