Baseline.html 13 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207
  1. <!DOCTYPE html>
  2. <html lang="en">
  3. <head>
  4. <meta charset="utf-8" />
  5. <meta name="viewport" content="width=device-width, initial-scale=1.0" /><meta name="generator" content="Docutils 0.18.1: http://docutils.sourceforge.net/" />
  6. <!-- Licensed under the Apache 2.0 License -->
  7. <link rel="stylesheet" type="text/css" href="_static/fonts/open-sans/stylesheet.css" />
  8. <!-- Licensed under the SIL Open Font License -->
  9. <link rel="stylesheet" type="text/css" href="_static/fonts/source-serif-pro/source-serif-pro.css" />
  10. <link rel="stylesheet" type="text/css" href="_static/css/bootstrap.min.css" />
  11. <link rel="stylesheet" type="text/css" href="_static/css/bootstrap-theme.min.css" />
  12. <meta name="viewport" content="width=device-width, initial-scale=1.0">
  13. <title>Baseline &#8212; MULTI-PARTY MEETING TRANSCRIPTION CHALLENGE 2.0</title>
  14. <link rel="stylesheet" type="text/css" href="_static/pygments.css" />
  15. <link rel="stylesheet" type="text/css" href="_static/guzzle.css" />
  16. <script data-url_root="./" id="documentation_options" src="_static/documentation_options.js"></script>
  17. <script src="_static/jquery.js"></script>
  18. <script src="_static/underscore.js"></script>
  19. <script src="_static/_sphinx_javascript_frameworks_compat.js"></script>
  20. <script src="_static/doctools.js"></script>
  21. <script src="_static/sphinx_highlight.js"></script>
  22. <link rel="index" title="Index" href="genindex.html" />
  23. <link rel="search" title="Search" href="search.html" />
  24. <link rel="next" title="Rules" href="Rules.html" />
  25. <link rel="prev" title="Track &amp; Evaluation" href="Track_setting_and_evaluation.html" />
  26. </head><body>
  27. <div class="related" role="navigation" aria-label="related navigation">
  28. <h3>Navigation</h3>
  29. <ul>
  30. <li class="right" style="margin-right: 10px">
  31. <a href="genindex.html" title="General Index"
  32. accesskey="I">index</a></li>
  33. <li class="right" >
  34. <a href="Rules.html" title="Rules"
  35. accesskey="N">next</a> |</li>
  36. <li class="right" >
  37. <a href="Track_setting_and_evaluation.html" title="Track &amp; Evaluation"
  38. accesskey="P">previous</a> |</li>
  39. <li class="nav-item nav-item-0"><a href="index.html">MULTI-PARTY MEETING TRANSCRIPTION CHALLENGE 2.0</a> &#187;</li>
  40. <li class="nav-item nav-item-this"><a href="">Baseline</a></li>
  41. </ul>
  42. </div>
  43. <div class="container-wrapper">
  44. <div id="mobile-toggle">
  45. <a href="#"><span class="glyphicon glyphicon-align-justify" aria-hidden="true"></span></a>
  46. </div>
  47. <div id="left-column">
  48. <div class="sphinxsidebar"><a href="
  49. index.html" class="text-logo">MULTI-PARTY MEETING TRANSCRIPTION CHALLENGE 2.0</a>
  50. <div class="sidebar-block">
  51. <div class="sidebar-wrapper">
  52. <div id="main-search">
  53. <form class="form-inline" action="search.html" method="GET" role="form">
  54. <div class="input-group">
  55. <input name="q" type="text" class="form-control" placeholder="Search...">
  56. </div>
  57. <input type="hidden" name="check_keywords" value="yes" />
  58. <input type="hidden" name="area" value="default" />
  59. </form>
  60. </div>
  61. </div>
  62. </div>
  63. <div class="sidebar-block">
  64. <div class="sidebar-toc">
  65. <p class="caption" role="heading"><span class="caption-text">Contents:</span></p>
  66. <ul class="current">
  67. <li class="toctree-l1"><a class="reference internal" href="Introduction.html">Introduction</a><ul>
  68. <li class="toctree-l2"><a class="reference internal" href="Introduction.html#call-for-participation">Call for participation</a></li>
  69. <li class="toctree-l2"><a class="reference internal" href="Introduction.html#timeline-aoe-time">Timeline(AOE Time)</a></li>
  70. <li class="toctree-l2"><a class="reference internal" href="Introduction.html#guidelines">Guidelines</a></li>
  71. </ul>
  72. </li>
  73. <li class="toctree-l1"><a class="reference internal" href="Dataset.html">Datasets</a><ul>
  74. <li class="toctree-l2"><a class="reference internal" href="Dataset.html#overview-of-training-data">Overview of training data</a></li>
  75. <li class="toctree-l2"><a class="reference internal" href="Dataset.html#detail-of-alimeeting-corpus">Detail of AliMeeting corpus</a></li>
  76. <li class="toctree-l2"><a class="reference internal" href="Dataset.html#get-the-data">Get the data</a></li>
  77. </ul>
  78. </li>
  79. <li class="toctree-l1"><a class="reference internal" href="Track_setting_and_evaluation.html">Track &amp; Evaluation</a><ul>
  80. <li class="toctree-l2"><a class="reference internal" href="Track_setting_and_evaluation.html#speaker-attributed-asr">Speaker-Attributed ASR</a></li>
  81. <li class="toctree-l2"><a class="reference internal" href="Track_setting_and_evaluation.html#evaluation-metric">Evaluation metric</a></li>
  82. <li class="toctree-l2"><a class="reference internal" href="Track_setting_and_evaluation.html#sub-track-arrangement">Sub-track arrangement</a></li>
  83. </ul>
  84. </li>
  85. <li class="toctree-l1 current"><a class="current reference internal" href="#">Baseline</a><ul>
  86. <li class="toctree-l2"><a class="reference internal" href="#overview">Overview</a></li>
  87. <li class="toctree-l2"><a class="reference internal" href="#quick-start">Quick start</a></li>
  88. <li class="toctree-l2"><a class="reference internal" href="#baseline-results">Baseline results</a></li>
  89. </ul>
  90. </li>
  91. <li class="toctree-l1"><a class="reference internal" href="Rules.html">Rules</a></li>
  92. <li class="toctree-l1"><a class="reference internal" href="Organizers.html">Organizers</a></li>
  93. <li class="toctree-l1"><a class="reference internal" href="Contact.html">Contact</a></li>
  94. </ul>
  95. </div>
  96. </div>
  97. </div>
  98. </div>
  99. <div id="right-column">
  100. <div role="navigation" aria-label="breadcrumbs navigation">
  101. <ol class="breadcrumb">
  102. <li><a href="index.html">Docs</a></li>
  103. <li>Baseline</li>
  104. </ol>
  105. </div>
  106. <div class="document clearer body">
  107. <section id="baseline">
  108. <h1>Baseline<a class="headerlink" href="#baseline" title="Permalink to this heading">¶</a></h1>
  109. <section id="overview">
  110. <h2>Overview<a class="headerlink" href="#overview" title="Permalink to this heading">¶</a></h2>
  111. <p>We will release an E2E SA-ASR baseline conducted on <a class="reference external" href="https://github.com/alibaba-damo-academy/FunASR">FunASR</a> at the time according to the timeline. The model architecture is shown in Figure 3. The SpeakerEncoder is initialized with a pre-trained speaker verification model from ModelScope. This speaker verification model is also be used to extract the speaker embedding in the speaker profile.</p>
  112. <p><img alt="model archietecture" src="_images/sa_asr_arch.png" /></p>
  113. </section>
  114. <section id="quick-start">
  115. <h2>Quick start<a class="headerlink" href="#quick-start" title="Permalink to this heading">¶</a></h2>
  116. <p>To run the baseline, first you need to install FunASR and ModelScope. (<a class="reference external" href="https://alibaba-damo-academy.github.io/FunASR/en/installation.html">installation</a>)<br />
  117. There are two startup scripts, <code class="docutils literal notranslate"><span class="pre">run.sh</span></code> for training and evaluating on the old eval and test sets, and <code class="docutils literal notranslate"><span class="pre">run_m2met_2023_infer.sh</span></code> for inference on the new test set of the Multi-Channel Multi-Party Meeting Transcription 2.0 (<a class="reference external" href="https://alibaba-damo-academy.github.io/FunASR/m2met2/index.html">M2MeT2.0</a>) Challenge.<br />
  118. Before running <code class="docutils literal notranslate"><span class="pre">run.sh</span></code>, you must manually download and unpack the <a class="reference external" href="http://www.openslr.org/119/">AliMeeting</a> corpus and place it in the <code class="docutils literal notranslate"><span class="pre">./dataset</span></code> directory:</p>
  119. <div class="highlight-shell notranslate"><div class="highlight"><pre><span></span>dataset
  120. <span class="p">|</span>——<span class="w"> </span>Eval_Ali_far
  121. <span class="p">|</span>——<span class="w"> </span>Eval_Ali_near
  122. <span class="p">|</span>——<span class="w"> </span>Test_Ali_far
  123. <span class="p">|</span>——<span class="w"> </span>Test_Ali_near
  124. <span class="p">|</span>——<span class="w"> </span>Train_Ali_far
  125. <span class="p">|</span>——<span class="w"> </span>Train_Ali_near
  126. Before<span class="w"> </span>running<span class="w"> </span><span class="sb">`</span>run_m2met_2023_infer.sh<span class="sb">`</span>,<span class="w"> </span>you<span class="w"> </span>need<span class="w"> </span>to<span class="w"> </span>place<span class="w"> </span>the<span class="w"> </span>new<span class="w"> </span><span class="nb">test</span><span class="w"> </span><span class="nb">set</span><span class="w"> </span><span class="sb">`</span>Test_2023_Ali_far<span class="sb">`</span><span class="w"> </span><span class="o">(</span>to<span class="w"> </span>be<span class="w"> </span>released<span class="w"> </span>after<span class="w"> </span>the<span class="w"> </span>challenge<span class="w"> </span>starts<span class="o">)</span><span class="w"> </span><span class="k">in</span><span class="w"> </span>the<span class="w"> </span><span class="sb">`</span>./dataset<span class="sb">`</span><span class="w"> </span>directory,<span class="w"> </span>which<span class="w"> </span>contains<span class="w"> </span>only<span class="w"> </span>raw<span class="w"> </span>audios.<span class="w"> </span>Then<span class="w"> </span>put<span class="w"> </span>the<span class="w"> </span>given<span class="w"> </span><span class="sb">`</span>wav.scp<span class="sb">`</span>,<span class="w"> </span><span class="sb">`</span>wav_raw.scp<span class="sb">`</span>,<span class="w"> </span><span class="sb">`</span>segments<span class="sb">`</span>,<span class="w"> </span><span class="sb">`</span>utt2spk<span class="sb">`</span><span class="w"> </span>and<span class="w"> </span><span class="sb">`</span>spk2utt<span class="sb">`</span><span class="w"> </span><span class="k">in</span><span class="w"> </span>the<span class="w"> </span><span class="sb">`</span>./data/Test_2023_Ali_far<span class="sb">`</span><span class="w"> </span>directory.<span class="w"> </span>
  127. <span class="sb">```</span>shell
  128. data/Test_2023_Ali_far
  129. <span class="p">|</span>——<span class="w"> </span>wav.scp
  130. <span class="p">|</span>——<span class="w"> </span>wav_raw.scp
  131. <span class="p">|</span>——<span class="w"> </span>segments
  132. <span class="p">|</span>——<span class="w"> </span>utt2spk
  133. <span class="p">|</span>——<span class="w"> </span>spk2utt
  134. </pre></div>
  135. </div>
  136. <p>For more details you can see <a class="reference external" href="https://github.com/alibaba-damo-academy/FunASR/blob/main/egs/alimeeting/sa-asr/README.md">here</a></p>
  137. </section>
  138. <section id="baseline-results">
  139. <h2>Baseline results<a class="headerlink" href="#baseline-results" title="Permalink to this heading">¶</a></h2>
  140. <p>The results of the baseline system are shown in Table 3. The speaker profile adopts the oracle speaker embedding during training. However, due to the lack of oracle speaker label during evaluation, the speaker profile provided by an additional spectral clustering is used. Meanwhile, the results of using the oracle speaker profile on Eval and Test Set are also provided to show the impact of speaker profile accuracy.</p>
  141. <p><img alt="baseline result" src="_images/baseline_result.png" /></p>
  142. </section>
  143. </section>
  144. </div>
  145. <div class="footer-relations">
  146. <div class="pull-left">
  147. <a class="btn btn-default" href="Track_setting_and_evaluation.html" title="previous chapter (use the left arrow)">Track &amp; Evaluation</a>
  148. </div>
  149. <div class="pull-right">
  150. <a class="btn btn-default" href="Rules.html" title="next chapter (use the right arrow)">Rules</a>
  151. </div>
  152. </div>
  153. <div class="clearer"></div>
  154. </div>
  155. <div class="clearfix"></div>
  156. </div>
  157. <div class="related" role="navigation" aria-label="related navigation">
  158. <h3>Navigation</h3>
  159. <ul>
  160. <li class="right" style="margin-right: 10px">
  161. <a href="genindex.html" title="General Index"
  162. >index</a></li>
  163. <li class="right" >
  164. <a href="Rules.html" title="Rules"
  165. >next</a> |</li>
  166. <li class="right" >
  167. <a href="Track_setting_and_evaluation.html" title="Track &amp; Evaluation"
  168. >previous</a> |</li>
  169. <li class="nav-item nav-item-0"><a href="index.html">MULTI-PARTY MEETING TRANSCRIPTION CHALLENGE 2.0</a> &#187;</li>
  170. <li class="nav-item nav-item-this"><a href="">Baseline</a></li>
  171. </ul>
  172. </div>
  173. <script type="text/javascript">
  174. $("#mobile-toggle a").click(function () {
  175. $("#left-column").toggle();
  176. });
  177. </script>
  178. <script type="text/javascript" src="_static/js/bootstrap.js"></script>
  179. <div class="footer">
  180. &copy; Copyright 2023, Speech Lab, Alibaba Group; ASLP Group, Northwestern Polytechnical University. Created using <a href="http://sphinx.pocoo.org/">Sphinx</a>.
  181. </div>
  182. </body>
  183. </html>