Audiocite.net is a corpus of read French speech downloaded in November 2021 from the <a href="https://www.audiocite.net/">Audiocite.net website</a>.
<br>
With a total duration of 6682 hours of audio recording, this corpus is the result of the voluntary work of 130 speakers.
The metadata is divided into 4 .jsons files (all(100%), train(80%), dev(10%) and test(10%)) to be used in NLP models.
<br>
The corpus and its metadata were uploaded through a script distributing the information in a .csv file. The use of these audio and metadata files is intended for pre-trained speech models.

<li>
    <b>Overview of the Corpus :</b>
</li>
<br>
<table>
    <tr>
      <th>Speakers Gender*</th>
      <th>Number of Files</th>
      <th>Number of Speakers</th>
      <th>Total Duration</th>
      <th>Avg. Duration</th>
      <th>Min. Duration</th>
      <th>Max. Duration</th>
    </tr>
    <tr>
      <td>M</td>
      <td>19345</td>
      <td>51</td>
      <td>4127:34:45</td>
      <td>00:12:48</td>
      <td>00:15:01</td>
      <td>06:39:43</td>
    </tr>
    <tr>
      <td>F</td>
      <td>8261</td>
      <td>70</td>
      <td>2272:18:23</td>
      <td>00:16:30</td>
      <td>00:03:49</td>
      <td>02:46:06</td>
    </tr>
    <tr>
      <td>U</td>
      <td>879</td>
      <td>9</td>
      <td>282:07:09</td>
      <td>00:19:15</td>
      <td>00:00:09</td>
      <td>03:30:16</td>
    </tr>
    <tr>
      <td>Total</td>
      <td>28485</td>
      <td>130</td>
      <td>6682:00:18</td>
      <td>00:14:04</td>
      <td>00:00:09</td>
      <td>06:39:43</td>
    </tr>
  </table>
<br>
*Beware speaker gender has been guessed and should not be considered as ground truth (cf. README.md of the audiocite.net_0.zip)

</p>

<pre>
Please cite the website Audiocite.net (<a href="https://www.audiocite.net/"></a>https://www.audiocite.net/</a>) if you use this dataset for your own research.
</pre>