<div dir="ltr">Hi Sir,<div><br></div><div>I have done a preliminary survey on the topic and have come up with a few points.</div><div><br></div><div>As per the description of the project idea "<font color="#333333" face="Helvetica, arial, freesans, clean, sans-serif"><span style="line-height:20px">Develop a language model for speech processing by extending a freely available corpus" I have come up with:</span></font></div>
<div><font color="#333333" face="Helvetica, arial, freesans, clean, sans-serif"><span style="line-height:20px"><br></span></font></div><div><font color="#333333" face="Helvetica, arial, freesans, clean, sans-serif"><span style="line-height:20px">We can go with CMUSphinx to build the language model for Bengali.This can be done as shown in Reference [1].</span></font></div>
<div><font color="#333333" face="Helvetica, arial, freesans, clean, sans-serif"><span style="line-height:20px">Now one point is that CMUSphinx has laready been tried.To do something new we can use J</span><span style="line-height:20px">ulius as I dont think it has been tried with Bengali.It will be definitely something new.</span></font></div>
<div><font color="#333333" face="Helvetica, arial, freesans, clean, sans-serif"><span style="line-height:20px"><br></span></font></div><div style><font color="#333333" face="Helvetica, arial, freesans, clean, sans-serif"><span style="line-height:20px">Next the problem is gathering data to train our system.I have found out to 2 approaches to get data.One is to use the data available on the shruthi Bangla ASR site [2] or we can use the algorithm in this paper [3] to generate phonemes consonants etc.</span></font></div>
<div><font color="#333333" face="Helvetica, arial, freesans, clean, sans-serif"><span style="line-height:20px"><br></span></font></div><div style><font color="#333333" face="Helvetica, arial, freesans, clean, sans-serif"><span style="line-height:20px">Third the actual STT can be done as mentioned in Reference [1] with the guidance of paper in Reference [4].Methods to reduce the noise and hence improve accuracy can be thought of (I havent research on it still).</span></font></div>
<div><font color="#333333" face="Helvetica, arial, freesans, clean, sans-serif"><span style="line-height:20px"><br></span></font></div><div style><font color="#333333" face="Helvetica, arial, freesans, clean, sans-serif"><span style="line-height:20px">Also I was curious whether we can make a TTS system.I was looking up at Dhvani [5].They say the Bengali module needs a lot improvements [6].Using the large data we have if we train Dhvani to improve and recognize digits even a good TTS system can be obtained.
</span></font></div><div><font color="#333333" face="Helvetica, arial, freesans, clean, sans-serif"><span style="line-height:20px"><br></span></font></div><div style><font color="#333333" face="Helvetica, arial, freesans, clean, sans-serif"><span style="line-height:20px">Finally, a very complete and concise documentation with all source code, method of implementation can be released for STT and TTS or both, which can be used by others to develop a language model for any Indic script.The proof=of-concept as said, will be done in Bengali and demonstrated.</span></font></div>
<div style><font color="#333333" face="Helvetica, arial, freesans, clean, sans-serif"><span style="line-height:20px"><br></span></font></div><div style><font color="#333333" face="Helvetica, arial, freesans, clean, sans-serif"><span style="line-height:20px">Thank you for your patience to go through this rather long mail.Please suggest any new ideas/concepts wherein I can improve upon what I wrote in this mail and come with a basic draft of the final objective.</span></font></div>
<div style><font color="#333333" face="Helvetica, arial, freesans, clean, sans-serif"><span style="line-height:20px"><br></span></font></div><div style><font color="#333333" face="Helvetica, arial, freesans, clean, sans-serif"><span style="line-height:20px">Thanking you in anticipation,</span></font></div>
<div style><font color="#333333" face="Helvetica, arial, freesans, clean, sans-serif"><span style="line-height:20px">Atanu </span></font></div><div><font color="#333333" face="Helvetica, arial, freesans, clean, sans-serif"><span style="line-height:20px">References:</span></font></div>
<div><font color="#333333" face="Helvetica, arial, freesans, clean, sans-serif"><span style="line-height:20px">[1]: </span></font><a href="http://cmusphinx.sourceforge.net/wiki/tutoriallm" target="_blank">http://cmusphinx.sourceforge.net/wiki/tutoriallm</a></div>
<div><span style="color:rgb(51,51,51);font-family:Helvetica,arial,freesans,clean,sans-serif;line-height:20px">[2]: </span><a href="http://cse.iitkgp.ac.in/~pabitra/shruti_corpus.html">http://cse.iitkgp.ac.in/~pabitra/shruti_corpus.html</a><font color="#333333" face="Helvetica, arial, freesans, clean, sans-serif"><span style="line-height:20px"><br>
</span></font></div><div><span style="color:rgb(51,51,51);font-family:Helvetica,arial,freesans,clean,sans-serif;line-height:20px">[3]: </span><a href="http://cse.iitkgp.ac.in/~pabitra/paper/ialp.pdf">http://cse.iitkgp.ac.in/~pabitra/paper/ialp.pdf</a><span style="color:rgb(51,51,51);font-family:Helvetica,arial,freesans,clean,sans-serif;line-height:20px"><br>
</span></div><div><span style="color:rgb(51,51,51);font-family:Helvetica,arial,freesans,clean,sans-serif;line-height:20px">[4]: </span><a href="http://cse.iitkgp.ac.in/~pabitra/paper/ococosda11.pdf">http://cse.iitkgp.ac.in/~pabitra/paper/ococosda11.pdf</a><span style="color:rgb(51,51,51);font-family:Helvetica,arial,freesans,clean,sans-serif;line-height:20px"><br>
</span></div><div><span style="color:rgb(51,51,51);font-family:Helvetica,arial,freesans,clean,sans-serif;line-height:20px">[5]: </span><a href="http://dhvani.sourceforge.net/">http://dhvani.sourceforge.net/</a><span style="color:rgb(51,51,51);font-family:Helvetica,arial,freesans,clean,sans-serif;line-height:20px"><br>
</span></div><div>[6]: <a href="http://dhvani.sourceforge.net/doc/bengali.html">http://dhvani.sourceforge.net/doc/bengali.html</a></div><div><font color="#333333" face="Helvetica, arial, freesans, clean, sans-serif"><span style="line-height:20px"><br>
</span></font></div></div><div class="gmail_extra"><br><br><div class="gmail_quote">On Tue, Apr 9, 2013 at 5:19 PM, Sankarshan Mukhopadhyay <span dir="ltr"><<a href="mailto:sankarshan.mukhopadhyay@gmail.com" target="_blank">sankarshan.mukhopadhyay@gmail.com</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div class="im">On Tue, Apr 9, 2013 at 10:24 AM, Atanu Ghosh <<a href="mailto:atanu1991@gmail.com">atanu1991@gmail.com</a>> wrote:<br>
> After going through the project ideas I have decided to work on this one.<br>
><br>
> "Add a language model for speech recognition software for Bengali language"<br>
><br>
> I was looking up at CMUSphinx for the same.I would like to know if there is<br>
> any other software to look up at or any other important references.<br>
<br>
</div>At this point CMUSphinx and Julius are the two obvious approaches. Do<br>
look up Shruti Bangla ASR. There is a heap of papers presented around<br>
CMUSphinx and Bengali as an ASR system - it would be a good place to<br>
get started.<br>
<br>
<br>
--<br>
sankarshan mukhopadhyay<br>
<<a href="https://twitter.com/#!/sankarshan" target="_blank">https://twitter.com/#!/sankarshan</a>><br>
_______________________________________________<br>
Project-ideas mailing list<br>
<a href="mailto:Project-ideas@lists.ankur.org.in">Project-ideas@lists.ankur.org.in</a><br>
<a href="http://lists.ankur.org.in/listinfo.cgi/project-ideas-ankur.org.in" target="_blank">http://lists.ankur.org.in/listinfo.cgi/project-ideas-ankur.org.in</a><br>
</blockquote></div><br></div>