CSR-I (WSJ0) Complete

<h3>Introduction</h3><br>
<p>LDC93S6A - Complete CSR-I corpus <a href="http://catalog.ldc.upenn.edu/LDC93S6B" rel="nofollow">LDC93S6B</a> - CSR-I Sennheiser speech <a href="http://catalog.ldc.upenn.edu/LDC93S6C" rel="nofollow">LDC93S6C</a> - CSR-I other speech</p><br>
<p>During 1991, the DARPA Spoken Language Program initiated efforts to build a new corpus to support research on large-vocabulary Continuous Speech Recognition (CSR) systems.</p><br>
<p>The first two CSR Corpora consist primarily of read speech with texts drawn from a machine-readable corpus of Wall Street Journal news text and are thus often known as WSJ0 and WSJ1. (Later sections of the CSR set of corpora, however, will consist of read texts from other sources of North American business news and eventually from other news domains).</p><br>
<p>The texts to be read were selected to fall within either a 5,000-word or a 20,000-word subset of the WSJ text corpus. (See the documentation for details). Some spontaneous dictation is included in addition to the read speech. The dictation portion was collected using journalists who dictated hypothetical news articles.</p><br>
<p>Two microphones are used throughout: a close-talking Sennheiser HMD414 and a secondary microphone, which may vary. The corpora are thus offered in three configurations: the speech from the Sennheiser, the speech from the other microphone and the speech from both; all three sets include all transcriptions, tests, documentation, etc.</p><br>
<p>In general, transcriptions of the speech, test data from ARPA evaluations, scores achieved by various speech recognition systems and software used in scoring are included on separate discs from the waveform data.</p><br>
<h3>Samples</h3><br>
<p>Please listen to this <a href="desc/addenda/LDC93S6A.wav">audio sample</a>.</p></br>
Portions © 1987-1989 Dow Jones & Company, Inc., © 1992, 1993 Trustees of the University of Pennsylvania

<h3>简介</h3><br><p>LDC93S6A — 完整CSR-I语料库；<a href="http://catalog.ldc.upenn.edu/LDC93S6B" rel="nofollow">LDC93S6B</a> — CSR-I 森海塞尔（Sennheiser）语音数据；<a href="http://catalog.ldc.upenn.edu/LDC93S6C" rel="nofollow">LDC93S6C</a> — CSR-I 其他语音数据</p><br><p>1991年，美国国防高级研究计划局（Defense Advanced Research Projects Agency, DARPA）口语语言项目启动了新语音语料库的构建工作，以支撑大词汇量连续语音识别（Continuous Speech Recognition, CSR）系统的相关研究。</p><br><p>前两款CSR语料库主要以标准朗读语音为主，其文本源自机器可读的《华尔街日报》（Wall Street Journal, WSJ）新闻语料库，因此常被称为WSJ0与WSJ1。不过后续的CSR系列语料库将采用北美商业新闻其他来源的朗读文本，最终覆盖其他新闻领域。</p><br><p>待朗读的文本选自WSJ文本语料库的5000词或20000词子集（详细说明请参阅相关文档）。除标准朗读语音外，本语料库还包含部分即兴口述语音：该口述部分由记者朗读虚构新闻稿件采集得到。</p><br><p>本次语料库全程采用两款麦克风采集：一款为贴耳式森海塞尔（Sennheiser）HMD414麦克风，另一款为可更换的辅助麦克风。因此本语料库提供三种配置版本：仅包含森海塞尔采集的语音、仅包含辅助麦克风采集的语音，以及同时包含两者的语音；三个版本均包含完整的转录文本、测试数据与文档资料等。</p><br><p>通常而言，语音转录文本、美国高级研究计划局（Advanced Research Projects Agency, ARPA）评估测试数据、各类语音识别系统的测试得分以及评分所用软件，均与波形音频数据存储在不同的光盘中。</p><br><h3>示例</h3><br><p>请收听此<a href="desc/addenda/LDC93S6A.wav">音频示例</a>。</p><br>部分内容 © 1987-1989 道琼斯公司（Dow Jones & Company, Inc.），© 1992、1993 宾夕法尼亚大学理事会（Trustees of the University of Pennsylvania）

搜集汇总

背景与挑战

背景概述

CSR-I (WSJ0) Complete是一个包含141小时英语语音录音的数据集，主要用于大词汇量连续语音识别研究，包含123名说话者的朗读录音及完整的转录和语言模型。