[Student-projects] Reg : Language model and Acoustic model for Malayalam as GSOC Project

[karan singla]

Hello,
I am Karan, a research student of IIIT Hyderabad and I am working primarily
in Machine Translation and have also worked with pocket sphinx and at&t
watson to build ASR for Hindi and Bengali as a part of SEECAT Project

http://bridge.cbs.dk/platform/?q=SEECAT

We also worked on making ASR for malayalam by Bootstraping from Hindi, but
the results were not good. actually worse( like 18% word accuracy )

Do we have resources available ?
1. Phonetic Dictionary ( phones need to mapped to English Phones, if we are
using Sphinx)
2. We can train a basic model from the dictionary and get grapheme to
phoneme mappings for all the words in the language model too. Thus creating
a bigger phonetic dictionary.

It will be great, if i could get guidance from where to start with ?


Regards,
Karan Singla
LTRC, IIIT-Hyderabad
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.smc.org.in/pipermail/student-projects-smc.org.in/attachments/20140305/c7fcbfbb/attachment-0001.htm>