(1)   项目名称(中、英文)

RASC863-G2 ——863地方普通话语音语料库(第2批,6地语音库)

RASC863-G2 -- 863 Regional Accent Speech Corpusgroup2, 6 regional corpus

(2)   项目简述

RASC863是在国家863高技术项目支持下完成的地方普通话语音语料库。其中第一批四个地方的数据(上海、广州、重庆和厦门)已经于2004年完成并发布,本语料库(RASC863-G2)是该项目的第二批数据库,共有6个地方的数据(长沙、洛阳、南京、南昌、太原、温州)。

    1996年863语音识别数据库以朗读语体为主,考虑了语音的音段平衡。随着语音识别技术的发展,制作口音和口语化的语音库变得重要起来。所以,我们这次由国家863项目基金支持的RASC863项目,在收集地方普通话语音语料库时,突出了口语化的特点,加大了语料覆盖范围。

 

Funded by the National 863 High-Tech Program, we collected a speech corpus RASC863 with four representative regional accents, namely Chongqing, Shanghai, Guangzhou and Xiamen.

The national Speech Corpus for ASR of the year 1996, also funded by the National 863 High-Tech Program, was mainly of reading style, considering the phonetic balance between segments. However, as the development of speech technology, it becomes important to build accented and spontaneous speech corpora. So, this time in Project RASC863, we have emphasized particularly on coverage and spontaneity of speech in much more areas of daily life, while collecting speech data of Changsha, Luoyang, Nanchang, Nanjing, Taiyuan, Wenzhou accented Mandarin.

3)单位名称(中、英文)

中国社会科学院语言所北京建国门内大街5号   中国社会科学院语言所语音研究室邮政编码:100732电话:(+86-010-65237408 / 85195394

Phonetics Lab, Institute of Linguistics, Chinese Academy of Social Sciences

5 Jian Guo Men Nei Da Jie, Beijing 100732
CHINA
Phone: +86-10-6523 7408
Fax: +86-10-85195396
E-mail: yinzhg@cass.org.cn

4)开发时间

20041月至20069

Period of development:

Jan. 2003-Jun. 2004

 

5)规模

每个方言点的发音人为200人(100+100女),共1200人。各方言点发音人情况按照事先设计的年龄、性别和教育背景分布。(详情见“发音人规范”)

RASC863-G2包括自然口语部分(口语独白和常见问题回答)和朗读(语音平衡句子、常用口语句和信息通讯语句)部分。自然口语部分分为依据话题的口语独白和回答问题两个部分:口语独白部分是由发音人从我们事先准备的160个话题中任意选择一个,然后讲述3-5分钟相关内容;回答问题部分是由每个发音人回答23个常见问题。朗读语料部分包括经过挑选的语音平衡句子共1890余句、460个常用口语句以及100个面向信息通讯应用的数字、字母、短信内容的混合语料。(详情见“语料设计规范”)

RASC863-G2录音数据收集近距(距嘴角距离2-8CM)和中距(20-50CM)两路数据,每路数据采用16000hz采样、16位、双声道WAV格式存储。(详情请见“录音和存储规范”)

 

The corpus consists of spontaneous speech, read speech and selected dialectical words. For the spontaneous speech, each speaker was asked to select a topic himself or from our prepared topic sheet with a variety of 160 topics and then to give a 4-5 minute spontaneous speech on the topic. Besides, each speaker was asked to answer 23 questions spontaneously. The read speech consists of 1895 phonetically balanced sentences selected automatically, 460 sentences frequently used in daily life , and 100 info-compound sentences (digits, characters, SMS) . 1200 speakers (200 from each region; balanced in terms of the age, sex, and educational background) were recruited in the project.

 

6)定价

[待商定]

7)相关技术文档

《地方口音普通话语音库RASC863-G2技术报告中文版》

《地方口音普通话语音库RASC863-G2技术报告英文版》

《地方口音普通话语音库RASC863-G2录音存储规范》

《地方口音普通话语音库RASC863-G2发音人规范》

《地方口音普通话语音库RASC863-G2语料设计规范》

《地方口音普通话语音库RASC863-G2标注规范》

 

A technical report on RASC863 (Chinese version)

A technical report on RASC863 (English version)

The specifications of recording and storing for RASC863

The specifications of speakers for RASC863

The specifications of text design for RASC863

The specifications of annotation for RASC863

 

8)语音库样例

RASC863-G2语音库样例

Samples:

Please refer to Samples of RASC863