Articulation Index LSCP

<h3>Introduction</h3><br>
<p>Articulation Index LSCP was developed by researchers at <a href="http://www.lscp.net/index.php?lang=en">Laboratoire de Sciences Cognitives et Psycholinguistique (LSCP), Ecole Normale Sup&eacute;rieure</a>. It revises and enhances a subset of Articulation Index (AIC) (<a href="../../../LDC2005S22">LDC2005S22</a>), a corpus of persons speaking English syllables. Changes include the addition of forced alignment to sound files, time alignment of syllable utterances and format conversions.</p><br>
<p>AIC consists of 20 American English speakers (12 males, 8 females) pronouncing syllables, some of which form actual words, but most of which are nonsense syllables. All possible Consonant-Vowel (CV) and Vowel-Consonant (VC) combinations were recorded for each speaker twice, once in isolation and once within a carrier-sentence, for a total of 25768 recorded syllables.</p><br>
<h3>Data</h3><br>
<p>Articulation Index LSCP alters AIC in the following ways.</p><br>
<ol><br>
<li>Time-alignments for the onset and offset of each word and syllable were generated through forced-alignment with a standard HMM-GMM (Hidden Markov Model-Gaussian Mixture Model) ASR system.</li><br>
<li>The time-alignments for the beginning and end of the syllables (whether in isolation or within a carrier sentence) were manually adjusted. The time-alignments for the other words in carrier sentences were not manually adjusted.</li><br>
<li>The recordings of isolated syllables were cut according to the manual time-alignments to remove the silent portions at the beginning and end, and the time-alignments were altered to correspond to the cut recordings.</li><br>
<li>The file naming scheme was slightly altered for compatibility with the <a href="http://kaldi.sourceforge.net/">Kaldi speech recognition toolkit</a>.</li><br>
<li>AIC contains a wide-band (16 KHz, 16-bit PCM) and a narrow-band (8 KHz, 8 bit u-law) version of the recordings distributed in sphere format. The LSCP version contains the wide-band version only distributed as wave files.</li><br>
</ol><br>
<p>This release does not include certain AIC triphone recordings (CVC, CCV or VCC).</p><br>
<p>Audio data is presented as 16kHz 16-bit flac compressed .wav files. The flac compression was added for distribution, and documentation may refer to the files as .wav files.</p><br>
<h3>Samples</h3><br>
<p>Please listen to this <a href="desc/addenda/LDC2015S12.wav">audio sample</a>.</p><br>
<h3>Updates</h3><br>
<p>None at this time.</p></br>
Portions © 2015 Tomas Bergvelt, Anna Kolesnikov, Xuan-Nga Cao, Thomas Schatz, Emmanuel Dupoux, © 2005, 2015 Trustees of the University of Pennsylvania

<h3>引言</h3><br><p>发音指数LSCP（Articulation Index LSCP）由<a href="http://www.lscp.net/index.php?lang=en">认知科学与心理语言学实验室（Laboratoire de Sciences Cognitives et Psycholinguistique，缩写LSCP），巴黎高等师范学院</a>的研究人员开发。该数据集修订并优化了发音指数语料库（Articulation Index，缩写AIC，对应编号<a href="../../../LDC2005S22">LDC2005S22</a>）的部分内容，AIC是一套包含英语音节发音的语料库。本次修订与优化涵盖了为音频文件添加强制对齐、音节语音的时间对齐以及格式转换等内容。</p><br><p>AIC语料库包含20名美式英语使用者（12名男性，8名女性）的音节发音数据，其中部分音节为实际存在的单词，其余多数为无意义音节。针对每位发音者，所有可能的辅音-元音（CV）与元音-辅音（VC）组合均被录制两次：一次为单独发音，另一次嵌入载体句中，最终总录制音节数达25768个。</p><br><h3>数据说明</h3><br><p>本数据集对AIC语料库进行了如下修改与优化：</p><br><ol><br><li>借助标准隐马尔可夫模型-高斯混合模型（Hidden Markov Model-Gaussian Mixture Model）自动语音识别（ASR）系统完成强制对齐，生成每个单词与音节的起止时间对齐信息。</li><br><li>对音节（无论单独发音还是嵌入载体句）的起止时间对齐信息进行人工校准，而载体句中其余单词的时间对齐信息未做人工调整。</li><br><li>根据人工校准后的时间对齐信息裁剪单独发音的音节录音，去除首尾静音片段，并同步调整时间对齐信息以匹配裁剪后的音频文件。</li><br><li>调整文件命名规则，以适配<a href="http://kaldi.sourceforge.net/">卡尔迪（Kaldi）语音识别工具包</a>。</li><br><li>AIC语料库包含两种版本的录音：宽带版（16 kHz、16位PCM编码）与窄带版（8 kHz、8位μ律编码），均以Sphere格式分发。而本LSCP版本仅保留宽带版录音，并以.wav文件格式分发。</li><br></ol><br><p>本次发布未包含AIC语料库中的部分三音素录音（CVC、CCV或VCC组合）。</p><br><p>音频数据以16 kHz、16位FLAC压缩的.wav文件形式提供。本次分发时添加了FLAC压缩步骤，相关文档中可能将此类文件直接称为.wav文件。</p><br><h3>样本</h3><br><p>请收听此<a href="desc/addenda/LDC2015S12.wav">音频样本</a>。</p><br><h3>更新情况</h3><br><p>暂无更新记录。</p><br><p>部分内容 © 2015 Tomas Bergvelt、Anna Kolesnikov、Xuan-Nga Cao、Thomas Schatz、Emmanuel Dupoux，© 2005、2015 宾夕法尼亚大学托管委员会</p>

搜集汇总

背景与挑战

背景概述

Articulation Index LSCP是一个英语语音数据集，由20名说话者录制了25768个音节，涵盖所有可能的CV和VC组合，并进行了时间对齐和格式转换优化。该数据集是Articulation Index (AIC)的增强版本，专注于音节发音，适用于语言识别、建模和发音研究等应用。