Initial commit for the addition of TIMIT dataset

[vrindaprabhu]

Below points needs to be addressed:

• Creation of dummy dataset is failing
• Need to check on the data representation
• License is not creative commons. Copyright: Portions © 1993 Trustees of the University of Pennsylvania

Also the links (except the download) point to the ami corpus! ;-)

@patrickvonplaten Requesting your comments, will be happy to address them!

[vrindaprabhu]

@patrickvonplaten could you please review and help me close this PR?

[lhoestq]

Awesome, thanks for adding this one !

This is pretty good overall, good job :)
I left a few comments

[vrindaprabhu]

@lhoestq Thank you so much for your comments and for patiently reviewing the code. Have hopefully included all the suggested changes. Let me know if any more changes are required.

Sorry the code had lots of silly errors from my side!:' Will be more careful from next time! :)

[lhoestq]

Thanks for the changes ! :)

added my last comments

Also it looks like the dummy data file is quite big (5MB), could you try to reduce its size a bit ?
To do so feel free to only keep 1 or 2 lines in the csv files (with is audio to TRUE and is_converted_audio to FALSE) and only keep the relevant wav/txt/phn files.

The idea is to have a very small set of data in order to test the script quickly and keeping the repository lightweight :)

[patrickvonplaten]

Looks good to me :-)

[lhoestq]

Thanks for the changes ! :)