CHiME2 Grid

<h3>Introduction</h3><br>
<p>CHiME2 Grid was developed as part of <a href="http://spandh.dcs.shef.ac.uk/chime_challenge/chime2013/">The 2nd CHiME Speech Separation and Recognition Challenge</a> and contains approximately 120 hours of English speech from a noisy living room environment. The CHiME Challenges focus on distant-microphone automatic speech recognition (ASR) in real-world environments.</p><br>
<p>CHiME2 Grid reflects the <a href="http://spandh.dcs.shef.ac.uk/chime_challenge/chime2013/chime2_task1.html">small vocabulary track</a> of the CHiME2 Challenge. The target utterances were taken from the <a href="http://spandh.dcs.shef.ac.uk/gridcorpus/">Grid corpus</a> and consist of 34 speakers reading simple 6-word sequences.</p><br>
<p>LDC also released CHiME2 WSJ0 (<a href="../../../LDC2017S10">LDC2017S10</a>) and CHiME3 (<a href="../../../LDC2017S24">LDC2017S24</a>).</p><br>
<h3>Data</h3><br>
<p>Data is divided into training, development and test sets. All data is provided as 16 bit WAV files sampled at 16 kHz. The noisy utterances are provided both in isolated form and in embedded form. The latter either involve five seconds of background noise before and after the utterance (in the training set) or they are mixed in continuous five minute noise background recordings (in the development and test sets). Seven hours of noise background not part of the training set are also included. The data is accompanied by one annotation file per speaker that includes additional technical information.</p><br>
<p>Also included is a baseline Hidden Markov Model (HMM)-based speech recogniser and a scoring tool designed for the 2nd CHiME Challenge to allow users to obtain keyword recognition scores from formatted result files, perform recognition and score the challenge data, and estimate parameters of speaker dependent HMMs.</p><br>
<h3>Samples</h3><br>
<p>Please listen to the following samples:</p><br>
<ul><br>
<li><a href="desc/addenda/LDC2017S07.clean.wav">Clean</a></li><br>
<li><a href="desc/addenda/LDC2017S07.embed.wav">Embedded</a></li><br>
<li><a href="desc/addenda/LDC2017S07.isolate.wav">Isolated</a></li><br>
<li><a href="desc/addenda/LDC2017S07.reverb.wav">Reverberated</a></li><br>
</ul><br>
<h3>Updates</h3><br>
<p>None at this time.</p></br>
Portions © 2017 Inria Nancy - Grand Est, University of Sheffield, Mitsubishi Electric Research Labs, Fondazione Bruno Kessler, © 2017 Trustees of the University of Pennsylvania

<h3>简介</h3><br><p>CHiME2 Grid是为<a href="http://spandh.dcs.shef.ac.uk/chime_challenge/chime2013/">第二届CHiME语音分离与识别挑战赛</a>开发的数据集，包含约120小时来自嘈杂客厅环境的英语语音。CHiME系列挑战赛聚焦真实场景下的远场麦克风自动语音识别（ASR，Automatic Speech Recognition）。</p><br><p>CHiME2 Grid对应第二届CHiME挑战赛的<a href="http://spandh.dcs.shef.ac.uk/chime_challenge/chime2013/chime2_task1.html">小词汇量赛道</a>。其目标语音取自<a href="http://spandh.dcs.shef.ac.uk/gridcorpus/">Grid语料库</a>，涵盖34位说话人朗读的简单6词语句序列。</p><br><p>LDC还发布了CHiME2 WSJ0（<a href="../../../LDC2017S10">LDC2017S10</a>）与CHiME3（<a href="../../../LDC2017S24">LDC2017S24</a>）。</p><br><h3>数据说明</h3><br><p>数据集划分为训练集、开发集与测试集。所有数据均以16 kHz采样的16位WAV格式文件提供。带噪语音同时提供孤立语音与嵌入语音两种形式：训练集的嵌入语音在语音前后各附带5秒背景噪声，而开发集与测试集的嵌入语音则混合了连续5分钟的背景噪声录音。此外还提供了不属于训练集的7小时背景噪声数据。每份数据均附带每位说话人的标注文件，包含额外技术信息。</p><br><p>本次发布还包含基于隐马尔可夫模型（Hidden Markov Model, HMM）的基线语音识别器，以及专为第二届CHiME挑战赛设计的评分工具，支持用户从格式化结果文件中获取关键词识别分数、完成语音识别并对挑战赛数据进行评分，以及估算与说话人相关的HMM参数。</p><br><h3>示例样本</h3><br><p>请收听以下示例：</p><br><ul><br><li><a href="desc/addenda/LDC2017S07.clean.wav">纯净语音</a></li><br><li><a href="desc/addenda/LDC2017S07.embed.wav">嵌入语音</a></li><br><li><a href="desc/addenda/LDC2017S07.isolate.wav">孤立语音</a></li><br><li><a href="desc/addenda/LDC2017S07.reverb.wav">混响语音</a></li><br></ul><br><h3>更新记录</h3><br><p>暂无更新。</p><br>部分内容 © 2017 法国国家信息与自动化研究所南希-大东部分部、谢菲尔德大学、三菱电机研究实验室、布鲁诺·凯塞勒基金会，以及 © 2017 宾夕法尼亚大学托管委员会