Kathbath asr

[bloodraven66]

What?

asr recipe for KathBath corpus - https://ai4bharat.iitm.ac.in/indic-superb

Todo-

• Train mr model
• change ffmpeg usage
• improve results on some languages
• Upload all models

[sw005320]

1. Can you add a result in README.md with pre-trained model links?
2. Can you also add the corpus information here? https://github.com/espnet/espnet/blob/master/egs2/README.md

[codecov]

Codecov Report

Merging #5369 (8cb9055) into master (88050b2) will increase coverage by 5.21%.
Report is 13 commits behind head on master.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master    #5369      +/-   ##
==========================================
+ Coverage   71.97%   77.18%   +5.21%     
==========================================
  Files         679      679              
  Lines       61703    61703              
==========================================
+ Hits        44410    47628    +3218     
+ Misses      17293    14075    -3218     

Flag	Coverage Δ
test_configuration_espnet2	∅ <ø> (∅)
test_integration_espnet1	65.73% <ø> (?)
test_integration_espnet2	48.53% <ø> (-0.02%)	⬇️
test_python_espnet1	20.26% <ø> (ø)
test_python_espnet2	52.08% <ø> (+0.05%)	⬆️
test_utils	23.10% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

see 60 files with indirect coverage changes

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more

[sw005320]

Please fix CI errors https://github.com/espnet/espnet/actions/runs/5754957584/job/15601353468

[mergify]

This pull request is now in conflict :(

[bloodraven66]

LGTM! Please let us know if you have finished your action items and then we can merge it~ Thanks for your contribution.

Sure, should be able to complete in 2 days, will update

[bloodraven66]

LGTM! Please let us know if you have finished your action items and then we can merge it~ Thanks for your contribution.

It is complete

[sw005320]

Can you fix the conflict?
Also, the README.md looks good, but is it possible to make it more similar (consistent) to the other README.md?

[bloodraven66]

Can you fix the conflict? Also, the README.md looks good, but is it possible to make it more similar (consistent) to the other README.md?

I did not keep the standard readme since there are 12 languages, 4 test set for each language, 12 * 4 * 2 rows for WER and CER. Is that ok?

[sw005320]

Can you fix the conflict? Also, the README.md looks good, but is it possible to make it more similar (consistent) to the other README.md?

I did not keep the standard readme since there are 12 languages, 4 test set for each language, 12 * 4 * 2 rows for WER and CER. Is that ok?

Sure, but how about adding the other information, e.g., method name, environment, etc. like https://github.com/espnet/espnet/blob/master/egs2/librispeech_100/asr1/README.md#e-branchformer

[bloodraven66]

Can you fix the conflict? Also, the README.md looks good, but is it possible to make it more similar (consistent) to the other README.md?

I did not keep the standard readme since there are 12 languages, 4 test set for each language, 12 * 4 * 2 rows for WER and CER. Is that ok?

Sure, but how about adding the other information, e.g., method name, environment, etc. like https://github.com/espnet/espnet/blob/master/egs2/librispeech_100/asr1/README.md#e-branchformer

Added

[sw005320]

Thanks a lot, @bloodraven66!