摘        要

语音库的制作和分发应该是一个系统工程，每一步都应该遵从特定的规范，以便数据交换。本文按照数据库制作规范，介绍863四大方言普通话语音语料库（包括上海、广州、重庆和厦门普通话）--RASC863 （863 annotated 4 regional accent speech corpus）。RASC863包括自然口语部分、朗读部分（语音平衡和常用口语句）和方言词汇。口语设立了160个话题，由发音人任意选择一个，然后讲述相关内容4-5分钟。朗读语料是挑选的语音平衡的句子共2200句和600个常用口语句。每个方言点的发音人为200人，共800人。覆盖不同年龄、性别和教育背景。

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-autospace:none'>

   2．语音库制作过程和一般规范

normal align=left style='text-align:left;text-indent:18.0pt; mso-char-indent-count:2.0'>具体涉及的制作规范和含义如表1所示。

normal align=center style='margin-left:9.0pt;text-align:center'>表1： 制作语音语料库的一般规范

normal style='text-indent:21.0pt;mso-char-indent-count:2.0; text-autospace:none'> 

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal align=center style='text-align:center;text-indent:18.0pt; mso-char-indent-count:2.0;text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal align=center style='text-align:center;text-indent:18.0pt; mso-char-indent-count:2.0;text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:22.1pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal align=left style='text-align:left;text-autospace:none'> 

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='margin-left:18.0pt;text-indent:-18.0pt;mso-list:l16 level1 lfo29; tab-stops:list 18.0pt;text-autospace:none'>

normal style='margin-left:18.0pt;text-indent:-18.0pt;mso-list:l16 level1 lfo29; tab-stops:list 18.0pt;text-autospace:none'>

normal style='margin-left:18.0pt;text-indent:-18.0pt;mso-list:l16 level1 lfo29; tab-stops:list 18.0pt;text-autospace:none'>

normal style='margin-left:18.0pt;text-indent:-18.0pt;mso-list:l16 level1 lfo29; tab-stops:list 18.0pt;text-autospace:none'>

normal style='margin-left:18.0pt;text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='text-indent:18.0pt;mso-char-indent-count:2.0; text-autospace:none'>

normal style='margin-left:10.75pt;mso-para-margin-left:-1.0gd; text-indent:-10.75pt;mso-text-indent-alt:-21.25pt'>       祖漪清，1998，连续语音数据库设计的科学性问题，《语音研究报告》，中国社会科学院。

normal style='margin-left:10.75pt;mso-para-margin-left:-1.0gd; text-indent:-10.75pt;mso-text-indent-alt:-21.25pt'>       [2] 王天庆，李爱军，“连续汉语语音识别语料库的设计”，第六届全国现代语音学学术会议，天津，2003年10月。

normal style='margin-left:10.75pt;mso-para-margin-left:-1.0gd; text-indent:-10.75pt;mso-text-indent-alt:-21.25pt'>       [3] , , “production and Validation of Speech  , , 2003.

normal style='margin-left:18.0pt;text-indent:-18.0pt;mso-char-indent-count: -2.0;text-autospace:none'>[4] Aijun Li, Chinese Prosody and Prosodic Labeling of Spontaneous Speech, Speech Prosody 2002, Aix-en-Provence.

normal style='margin-left:18.0pt;text-indent:-18.0pt;mso-char-indent-count: -2.0;text-autospace:none'>[5] Xiaoxia Chen, Aijun Li, et. al.  Application of SAMPA-C in SC, Beijing.

normal style='margin-left:18.0pt;text-indent:-18.0pt;mso-char-indent-count: -2.0;text-autospace:none'>[6] 李荣主编，汉语方言词典系列，上海、广州、贵州和厦门方言方言辞典，中国社会科学院语言所。

normal style='margin-left:13.5pt;text-indent:-13.5pt;mso-char-indent-count: -1.5'>[7] Aijun Li Wang, A Contrastive Investigation of Standard Mandarin and Accented Mandarin, Eurospeech2003.

normal style='margin-left:22.5pt;text-indent:-22.5pt;mso-char-indent-count: -2.5'>

normal style='margin-left:18.0pt;text-indent:-18.0pt;mso-char-indent-count: -2.0'>

normal style='margin-left:18.0pt;text-indent:-18.0pt;mso-char-indent-count: -2.0;text-autospace:none'>[10] 陈娟文、

normal style='margin-left:18.0pt;text-indent:-18.0pt;mso-char-indent-count: -2.0;text-autospace:none'>[11] 陈娟文、