Skip to content

GitLab

Switch branch/tag
  1. 19 Jan, 2023 2 commits
    • Zhaoheng Ni's avatar
      Add modularized SSL training recipe (#2876) · 2eaefe27
      Zhaoheng Ni authored 
      Summary:
TorchAudio currently has one training recipe for HuBET + LibriSpeech pre-training. It may not suit well when users want to use customized dataset, or use a new training objective (such as contrastive loss in Wav2Vec2). The PR addresses the issue by providing a modularized training recipe for audio self-supervised learning. Users can inject customized model module, loss function, optimizer, lr scheduler, and datamodule for training a SSL model.

Pull Request resolved: https://github.com/pytorch/audio/pull/2876

Reviewed By: hwangjeff

Differential Revision: D42617414

Pulled By: nateanl

fbshipit-source-id: 6413df45a9d106ed1d5ff830bf628c54368c5792
      2eaefe27
    • hwangjeff's avatar
      Simplify train step in Conformer RNN-T LibriSpeech recipe (#2981) · c6a52355
      hwangjeff authored 
      Summary:
In the Conformer RNN-T LibriSpeech recipe, there's no need to perform manual optimization. This PR modifies the recipe to use automatic optimization instead.

Pull Request resolved: https://github.com/pytorch/audio/pull/2981

Reviewed By: mthrok

Differential Revision: D42507228

Pulled By: hwangjeff

fbshipit-source-id: 9712add951eba356e39f7e8c8dc3bf584ba48309
      c6a52355
  3. 17 Jan, 2023 1 commit
  5. 16 Jan, 2023 1 commit
    • Robin Scheibler's avatar
      Fixes examples/source_separation for WSJ0_2mix dataset (#2987) · f9d38796
      Robin Scheibler authored 
      Summary:
The `examples/source_separation` scripts use inconsistent keyword to indicate the WSJ0_2mix dataset. This PR does the following.

1. Use `wsj0mix` consistently as keyword indicating the WSJ0_2mix dataset
2. Corrects `args.data_dir` to `args.root_dir` in eval.py
3. Modify the parameters of `pytorch_lightning.Trainer` according to latest version (use `accelerator="gpu"` and `devices=args.num_devices`, instead of just `gpus=args.num_devices`)

Pull Request resolved: https://github.com/pytorch/audio/pull/2987

Reviewed By: xiaohui-zhang

Differential Revision: D42536992

Pulled By: nateanl

fbshipit-source-id: 10a80263ad7054b1629d8fa023676b607e633d76
      f9d38796
  7. 13 Jan, 2023 1 commit
  9. 30 Dec, 2022 1 commit
  11. 17 Dec, 2022 1 commit
  13. 16 Dec, 2022 1 commit
    • Caroline Chen's avatar
      Rename resampling_method options (#2922) · e6bebe6a
      Caroline Chen authored 
      Summary:
resolves https://github.com/pytorch/audio/issues/2891

Rename `resampling_method` options to more accurately describe what is happening. Previously the methods were set to `sinc_interpolation` and `kaiser_window`, which can be confusing as both options actually use sinc interpolation methodology, but differ in the window function used. As a result, rename `sinc_interpolation` to `sinc_interp_hann` and `kaiser_window` to `sinc_interp_kaiser`. Using an old option will throw a warning, and those options will be deprecated in 2 released. The numerical behavior is unchanged.

Pull Request resolved: https://github.com/pytorch/audio/pull/2922

Reviewed By: mthrok

Differential Revision: D42083619

Pulled By: carolineechen

fbshipit-source-id: 9a9a7ea2d2daeadc02d53dddfd26afe249459e70
      e6bebe6a
  15. 29 Nov, 2022 1 commit
  17. 28 Nov, 2022 1 commit
  19. 17 Nov, 2022 1 commit
  21. 16 Nov, 2022 2 commits
  23. 17 Oct, 2022 1 commit
  25. 14 Oct, 2022 2 commits
  27. 13 Oct, 2022 2 commits
  29. 12 Oct, 2022 2 commits
    • Nikita Shulga's avatar
      Fix typos in tacotron2 tutorial (#2761) · 7aabcbd4
      Nikita Shulga authored 
      Summary:
`publishe`->`published`

Also, not sure if it should be `pre-trained weight is published` or `pre-trained weights are published`

Pull Request resolved: https://github.com/pytorch/audio/pull/2761

Reviewed By: carolineechen

Differential Revision: D40313042

Pulled By: malfet

fbshipit-source-id: c22085ca0b1125a06aa04bf38231d0a9fbfed00b
      7aabcbd4
    • Zhaoheng Ni's avatar
      Improve hubert recipe for pre-training and fine-tuning (#2744) · 27433050
      Zhaoheng Ni authored 
      Summary:
following pr https://github.com/pytorch/audio/issues/2716
- For preprocessing
  - The HuBERT feature takes lots of memory which may not fit some machines. Enable to use a subset of feature for training a k-means model.

- For pre-training
  - Normalize the loss based on the total number of masked frames across all GPUs.
  - Use mixed precision training. fp16 is not well supported in pytorch_lightning.
  - Log accuracies of masked/unmasked frames during training.
  - Clip the gradients with norm `10.0`.

- For ASR fine-tuning
  - Normalize the loss based on the total number of batches across all GPUs, same as in the conformer recipe of TorchAudio.
  - Use mixed precision training.
  - Add "|" after the end of transcription to capture the silence/word termination, same as in fairseq recipe.

- Update the WER results on LibriSpeech dev and test sets.

|                   | WER% (Viterbi)|  WER% (KenLM) |
|:-----------------:|--------------:|--------------:|
| dev-clean         |       10.9    |       4.2     |
| dev-other         |       17.5    |       9.4     |
| test-clean        |       10.9    |       4.4     |
| test-other        |       17.8    |       9.5     |

Pull Request resolved: https://github.com/pytorch/audio/pull/2744

Reviewed By: carolineechen

Differential Revision: D40282322

Pulled By: nateanl

fbshipit-source-id: 4723584c912e70e8970149fe09de005385eaab90
      27433050
  31. 07 Oct, 2022 1 commit
  33. 06 Oct, 2022 1 commit
  35. 05 Oct, 2022 1 commit
  37. 03 Oct, 2022 1 commit
  39. 23 Sep, 2022 2 commits
  41. 22 Sep, 2022 2 commits
  43. 21 Sep, 2022 2 commits
  45. 14 Sep, 2022 1 commit
  47. 13 Sep, 2022 1 commit
  49. 09 Sep, 2022 1 commit
  51. 06 Sep, 2022 1 commit
  53. 26 Aug, 2022 1 commit
  55. 18 Aug, 2022 3 commits
    • moto's avatar
      Update ASR inference tutorial (#2631) · 189edb1b
      moto authored 
      Summary:
* Use download_asset
* Remove notes around nightly
* Print versions first
* Remove duplicated import

Pull Request resolved: https://github.com/pytorch/audio/pull/2631

Reviewed By: carolineechen

Differential Revision: D38830395

Pulled By: mthrok

fbshipit-source-id: c9259df33562defe249734d1ed074dac0fddc2f6
      189edb1b
    • moto's avatar
      Update notes around nightly build and third parties (#2632) · 55ce80b1
      moto authored 
      Summary:
Google Colab now has torchaudio 0.12 pre-installed.
This commit removes the note about nightly build.

Pull Request resolved: https://github.com/pytorch/audio/pull/2632

Reviewed By: carolineechen

Differential Revision: D38827632

Pulled By: mthrok

fbshipit-source-id: ac769780868b741c3012357d589ec0019d9af6eb
      55ce80b1
    • moto's avatar
      Tweak tutorials (#2630) · cab2bb44
      moto authored 
      Summary:
Resolves the following warnings

```
/torchaudio/docs/source/tutorials/asr_inference_with_ctc_decoder_tutorial.rst:195: WARNING: Unexpected indentation.
/torchaudio/docs/source/tutorials/asr_inference_with_ctc_decoder_tutorial.rst:446: WARNING: Unexpected indentation.
/torchaudio/docs/source/tutorials/audio_io_tutorial.rst:559: WARNING: Content block expected for the "note" directive; none found.
/torchaudio/docs/source/tutorials/mvdr_tutorial.rst:338: WARNING: Bullet list ends without a blank line; unexpected unindent.
```

Pull Request resolved: https://github.com/pytorch/audio/pull/2630

Reviewed By: nateanl

Differential Revision: D38816632

Pulled By: mthrok

fbshipit-source-id: 135ded4e064d136be67ce24439e96f5e9c9ce635
      cab2bb44
  57. 10 Aug, 2022 1 commit
  59. 05 Aug, 2022 1 commit
    • Caroline Chen's avatar
      Add note for lexicon free decoder output (#2603) · 33485b8c
      Caroline Chen authored 
      Summary:
``words`` field of CTCHypothesis is empty if no lexicon is provided, which produces confusing output (see issue https://github.com/pytorch/audio/issues/2584) when following our tutorial example with lexicon free usage. This PR adds a note in both docs and tutorial.

Followup: determine if we want to modify the behavior of ``words`` in the lexicon free case. One option is to merge and then split the generated tokens by the input silent token to populate the words field, but this is tricky since the meaning of a "word" in the lexicon free case can be vague and not all languages have whitespaces between words, etc

Pull Request resolved: https://github.com/pytorch/audio/pull/2603

Reviewed By: mthrok

Differential Revision: D38459709

Pulled By: carolineechen

fbshipit-source-id: d64ff186df4633f00e94c64afeaa6a50cebf2934
      33485b8c