Skip to main content
zenodoopen

VIVOS: Vietnamese Speech Corpus for ASR

<p><strong>VIVOS Corpus</strong></p> <p>VIVOS is a free Vietnamese speech corpus consisting of 15 hours of recording speech prepared for Automatic Speech Recognition task.</p> <p>The corpus was published by AILAB, a computer science lab of VNUHCM - University of Science, with&nbsp;<strong>Prof. Vu Hai Quan</strong>&nbsp;is the head of.</p> <p>We publish this corpus in hope to attract more scientists to solve Vietnamese speech recognition problems. The corpus should only be used for academic purposes.</p> <p><strong>License</strong></p> <p>Creative Commons Attribution NonCommercial ShareAlike v4.0 (CC BY-NC-SA 4.0) (<a href="https://creativecommons.org/licenses/by-nc-sa/4.0/legalcode">details</a>)</p> <p><strong>Associated paper</strong></p> <p>Please cite this paper when using VIVOS corpus for research</p> <p>&quot;A non-expert Kaldi recipe for Vietnamese Speech Recognition System&quot;, Hieu-Thi Luong and Hai-Quan Vu, in Proc. <em>WLSI/OIAF4HLT2016</em> (<a href="https://aclanthology.org/W16-5207/">paper</a>)</p> <p><strong>Contact</strong></p> <p><a href="mailto:ailab@hcmus.edu.vn">ailab@hcmus.edu.vn</a></p> <p>&nbsp;</p> <p><strong>Corpus properties</strong></p> <p>Speech was recorded in a quiet environment with high quality microphone, speakers were asked to read one sentence at a time.</p> <table> <thead> <tr> <th scope="col">&nbsp;</th> <th scope="col">Training</th> <th scope="col">Testing</th> </tr> </thead> <tbody> <tr> <td>Speakers</td> <td>46</td> <td>19</td> </tr> <tr> <td>Male</td> <td>22</td> <td>12</td> </tr> <tr> <td>Female</td> <td>24</td> <td>7</td> </tr> <tr> <td>Utterances</td> <td>11660</td> <td>760</td> </tr> <tr> <td>Duration</td> <td>14:55</td> <td>00:45</td> </tr> <tr> <td>Unique Syllables</td> <td>4617</td> <td>1692</td> </tr> </tbody> </table> <p><br> <strong>Evaluations</strong></p> <p>The corpus was evaluated using our non-expert recipe for Vietnamese Speech Recognition system which is described&nbsp;in the associated paper.</p> <table> <thead> <tr> <th scope="col">&nbsp;</th> <th scope="col">baseline</th> <th scope="col">+pitch</th> <th scope="col">+tone</th> </tr> </thead> <tbody> <tr> <td>mGMM</td> <td>19.66</td> <td>15.14</td> <td>14.91</td> </tr> <tr> <td>mGMM+MMI</td> <td>18.08</td> <td>14.96</td> <td>13.91</td> </tr> <tr> <td>mGMM+SAT</td> <td>15.79</td> <td>12.07</td> <td>12.13</td> </tr> <tr> <td>mDNN+SAT</td> <td>13.34</td> <td>9.54</td> <td>9.48</td> </tr> </tbody> </table> <p><strong>Notice</strong></p> <p>This is the official replacement for http://ailab.hcmus.edu.vn/vivos/</p> <p>&nbsp;</p>

ShareScore

28/100

Overall dataset sharing score

Score breakdown

These five areas show where the dataset supports — or may limit — practical reuse.

Stewardship
8
Harmonization
4
Access
12
Reuse readiness
0
Engagement
4

Topics