[BC-Breaking] Remove compute_kaldi_pitch #3368
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/audio/3368
Note: Links to docs will display an error until the docs builds have been completed. ❌ 5 New FailuresAs of commit 0db6e14: NEW FAILURES - The following jobs have failed:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
mthrok
force-pushed
the
remove-kaldi-integration
branch
2 times, most recently
from
eb4dcea to
b7a34e7
Compare
|
Also, could make sense to just create a GitHub feature request issue in torchaudio for some pitch extraction algos (which could be useful for TTS?), and later on add a new function reimplemented from the ground up without any kaldi-compat as a goal |
mthrok
force-pushed
the
remove-kaldi-integration
branch
from
b7a34e7 to
0102f69
Compare
Yeah, that makes sense, however I don't know much about what kind of pitch extraction ppl are using these days, so I don't know what are the good candidates. I also hear that people use NN-based pitch extraction. In that case, it requires more work to add such. |
This commit removes compute_kaldi_pitch function and the underlying Kaldi integration from torchaudio. Kaldi pitch function was added in a short period of time by integrating the original Kaldi implementation, instead of reimplementing it in PyTorch. The Kaldi integration employed a hack which replaces the base vector/matrix implementation of Kaldi with PyTorch Tensor so that there is only one blas library within torchaudio. Recently, we are making torchaudio more lean, and we don't see a wide adoption of kaldi_pitch feature, so we decided to remove them. See some of the discussion pytorch#1269
mthrok
force-pushed
the
remove-kaldi-integration
branch
from
0102f69 to
0db6e14
Compare
|
@mthrok has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator. |
Follow up of: pytorch#3368 Remove files and lines no longer used.
This commit removes compute_kaldi_pitch function and the underlying Kaldi integration from torchaudio.
Kaldi pitch function was added in a short period of time by integrating the original Kaldi implementation, instead of reimplementing it in PyTorch.
The Kaldi integration employed a hack which replaces the base vector/matrix implementation of Kaldi with PyTorch Tensor so that there is only one blas library within torchaudio.
Recently, we are making torchaudio more lean, and we don't see a wide adoption of kaldi_pitch feature, so we decided to remove them.
See some of the discussion #1269