<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=us-ascii">
</head>
<body>
<div data-ogsc="" style="">
<meta content="text/html; charset=us-ascii" data-ogsc="" style="">
</div>
<div dir="auto" style="direction: ltr; margin: 0px; padding: 0px; font-family: sans-serif; font-size: 11pt; color: black;">
<a href="https://www.aclweb.org/anthology/2020.lrec-1.789" data-ogsc="" style=""></a></div>
<div dir="auto" style="direction: ltr; margin: 0px; padding: 0px; font-family: sans-serif; font-size: 11pt; color: black; text-align: left;">
</div>
<div dir="auto" style="direction: ltr; margin: 0px; padding: 0px; font-family: sans-serif; font-size: 11pt; color: black; text-align: left;">
<span style="font-size: 11pt;">Hello,</span></div>
<div dir="auto" style="direction: ltr; margin: 0px; padding: 0px; font-family: sans-serif; font-size: 11pt; color: black; text-align: left;">
<span style="font-size: 11pt;"><br>
</span></div>
<div dir="auto" style="direction: ltr; margin: 0px; padding: 0px; font-family: sans-serif; font-size: 11pt; color: black; text-align: left;">
<span style="font-size: 11pt;">You might want to look into kaldi as Kavya suggested or deepspeech (Mozilla) which is open source with some active community. T</span><span style="font-size: 11pt;">raining is GPU heavy.</span></div>
<div dir="auto" style="direction: ltr; margin: 0px; padding: 0px; font-family: sans-serif; font-size: 11pt; color: black; text-align: left;">
<span style="font-size: 11pt;"><br>
</span></div>
<div dir="auto" style="direction: ltr; margin: 0px; padding: 0px; font-family: sans-serif; font-size: 11pt; color: black; text-align: left;">
<span style="font-size: 11pt;">The following publication provides 24 hour single speaker training data (in Malayalam and 2 additional languages), recorded in studio quality.</span></div>
<div dir="auto" style="direction: ltr; margin: 0px; padding: 0px; font-family: sans-serif; font-size: 11pt; color: black; text-align: left;">
</div>
<div dir="auto" style="direction: ltr; margin: 0px; padding: 0px; font-family: sans-serif; font-size: 11pt; color: black; text-align: left;">
<br>
</div>
<div dir="auto" style="direction: ltr; margin: 0px; padding: 0px; font-family: sans-serif; font-size: 11pt; color: black; text-align: left;">
<div dir="auto" style="margin: 0px; font-size: 11pt; color: black; text-align: left;">
<a href="https://www.aclweb.org/anthology/2020.lrec-1.789" data-ogsc="" style="">https://www.aclweb.org/anthology/2020.lrec-1.789</a>
<br>
<br>
</div>
<div dir="auto" style="margin: 0px; font-size: 11pt; color: black; text-align: left;">
<div dir="auto" style="margin: 0px; font-size: 11pt; color: black;">Experiments in paper are for TTS. The training data works both ways. Contact<span style="font-size: 11pt;"> the authors by email, they might have trained models for deepspeech or kaldi. I know
they have Hindi ASR (and TTS) models from an earlier <a href="https://cvit.iiit.ac.in/research/projects/cvit-projects/facetoface-translation" data-ogsc="" style="">
publication</a>, unsure if they have Malayalam model but I'm hoping it's in the works. They can possibly point you in the direction of better libraries and resources, as they're actively working in the space. There is active research (and some development)
going on as part of a project there.</span></div>
<div dir="auto" style="margin: 0px; font-size: 11pt; color: black;"><span style="font-size: 11pt;"><br>
</span></div>
<div dir="auto" style="margin: 0px; font-size: 11pt; color: black;"><span style="font-size: 11pt;">Disclaimer: I am in some capacity still part of this group.</span></div>
</div>
<br>
</div>
<div dir="auto" style="direction: ltr; margin: 0px; padding: 0px; font-family: sans-serif; font-size: 11pt; color: black;">
<span id="OutlookSignature">
<div dir="auto" style="direction: ltr; margin: 0px; padding: 0px; font-family: sans-serif; font-size: 11pt; color: black;">
--<br>
</div>
<div dir="auto" style="direction: ltr; margin: 0px; padding: 0px; font-family: sans-serif; font-size: 11pt; color: black;">
Jerin Philip</div>
</span><br>
</div>
<hr tabindex="-1" style="display: inline-block; width: 98%;" data-ogsc="">
<div id="divRplyFwdMsg" dir="ltr" data-ogsc="" style=""><font face="Calibri, sans-serif" color="#000000" style="font-size: 11pt;" data-ogsc=""><b>From:</b> discuss <discuss-bounces@lists.smc.org.in> on behalf of Kavya Manohar <sakhi.kavya@gmail.com><br>
<b>Sent:</b> Monday, June 22, 2020 7:39:22 AM<br>
<b>To:</b> Discussion list of Swathanthra Malayalam Computing <discuss@lists.smc.org.in><br>
<b>Subject:</b> Re: [smc-discuss] Speech to text tool for Malayalam</font>
<div> </div>
</div>
<div data-ogsc="" style="">
<div dir="ltr">
<div dir="ltr">
<div>Hi Julin,</div>
<div><br>
</div>
<div>We do not have a ready to use, open source speech to text tool for Malayalam as of now.</div>
<div><br>
</div>
<div>But there are works going on in this domain. I am working on a <a href="https://kaldi-asr.org/" data-ogsc="" style="">
Kaldi</a> based solution as an academic researcher. This <a href="https://gitlab.com/kavyamanohar/malayalam-spoken-digit-recognizer" data-ogsc="" style="">
repo </a>demonstrates the development of a spoken digit recognizer for Malayalam. A past attempt on Malayalam ASR using CMUSphinx by Sreenadh is available
<a href="https://github.com/sreecodeslayer/ml-am-lm-cmusphinx" data-ogsc="" style="">
here</a>.<br>
</div>
<div><br>
</div>
<div>SMC has taken initiatives for collecting <a href="https://msc.smc.org.in" data-ogsc="" style="">
Malayalam speech corpus</a> for training general purpose automatic speech recognition (ASR) systems. The audio recorded so far is available
<a href="https://gitlab.com/smc/msc" data-ogsc="" style="">here</a>.<br>
</div>
<div>A grapheme to phoneme converter, <a href="https://gitlab.com/smc/mlphon" data-ogsc="" style="">
mlphon</a> that creates a phonetic lexicon (a dictionary of Malayalam words and its pronunciation) needed for ASR training is another project. Mlphon is available as a
<a href="https://pypi.org/project/mlphon/" data-ogsc="" style="">Pypi</a> library.</div>
<div></div>
<div><br>
</div>
<div></div>
<div>Regards<br>
</div>
<div><br>
</div>
</div>
-- <br>
<div dir="ltr" class="x_gmail_signature">
<div dir="ltr">
<div>Kavya Manohar</div>
<div>Research Scholar</div>
<div>College of Engineering Trivandrum<br>
</div>
<a href="https://kavyamanohar.com" target="_blank" data-ogsc="" style="">https://kavyamanohar.com</a><br>
</div>
</div>
</div>
</div>
</body>
</html>