Skip to content

GitLab

Switch branch/tag
  1. 16 Feb, 2022 6 commits
    • Zhaoheng Ni's avatar
      Add complex dtype support in functional autograd test (#2244) · eeba91dc
      Zhaoheng Ni authored 
      Summary:
In autograd tests, to guarantee the precision, the dtype of Tensors are converted to `torch.float64` if they are real. However, the complex dtype is not considered. This PR adds `self.complex_dtype` support to the inputs.

Pull Request resolved: https://github.com/pytorch/audio/pull/2244

Reviewed By: mthrok

Differential Revision: D34272998

Pulled By: nateanl

fbshipit-source-id: e8698a74d7b8d99ee0fcb5f5cb5f2ffc8c80b9b5
      eeba91dc
    • Caroline Chen's avatar
      Fix lm used for ctc decoder example (#2235) · c2decba4
      Caroline Chen authored 
      Summary:
LM in example script was unintentionally changed to None when adding no LM support previously. this changes it back and is consistent with the WERs listed in the readme

Pull Request resolved: https://github.com/pytorch/audio/pull/2235

Reviewed By: nateanl

Differential Revision: D34273042

Pulled By: carolineechen

fbshipit-source-id: 824b1ce18195e39dc534b2ec9c5312bbe3bb1812
      c2decba4
    • Zhaoheng Ni's avatar
      Add shebang lines to scripts in emformer_rnnt recipes (#2237) · aac83fe5
      Zhaoheng Ni authored 
      Summary: Pull Request resolved: https://github.com/pytorch/audio/pull/2237

Reviewed By: mthrok

Differential Revision: D34267000

Pulled By: nateanl

fbshipit-source-id: 4c264aea6cf3fba5d8728d5fe60f9f471815852d
      aac83fe5
    • Zhaoheng Ni's avatar
      Add EMFORMER_RNNT_BASE_MUSTC bundle to torchaudio.prototype (#2241) · 99b5ef5c
      Zhaoheng Ni authored 
      Summary:
This PR provides a RNNTBundle that is pre-trained on the MuST-C release v2.0 dataset.
The model preserves the casing and punctuations of the transcripts when training the SentencePiece model.

Here is the model performance on the dev and test sets of MuST-C 2.0:
|                   |          WER |
|:-----------------:|-------------:|
| dev               |       0.190  |
| tst-COMMON        |       0.213  |
| tst-HE            |       0.186  |

Pull Request resolved: https://github.com/pytorch/audio/pull/2241

Reviewed By: mthrok

Differential Revision: D34267792

Pulled By: nateanl

fbshipit-source-id: 67bca9f277e66d41a4530d01615f249b3cec7167
      99b5ef5c
    • Zhaoheng Ni's avatar
      Refactor ArgumentParser arguments in emformer_rnnt recipes (#2236) · 81f56f64
      Zhaoheng Ni authored 
      Summary:
Replace underscore with dash in ArgumentParser's arguments.

Pull Request resolved: https://github.com/pytorch/audio/pull/2236

Reviewed By: mthrok

Differential Revision: D34266977

Pulled By: nateanl

fbshipit-source-id: ceacac12c04016a8dbf2a1a7d6bbcf65d4d53d21
      81f56f64
    • moto's avatar
      Fix prototype exclusion in release (#2225) · a007e922
      moto authored 
      Summary:
This commit fixes the feature to exclude `torchaudio.prototype` module.

In `setup.py` there is a special case that is triggered if the commit is on release branch or release tag, that  excludes `torchaudio.prototype`. This was introduced to make it easy for release-related work.
It turned out that the submodules under `torchaudio.prototype`, such as `torchaudio.prototype.pipelines`, are not properly excluded from packaging.
These sub modules did not exist in previous releases, so it was not an issue.

**Note** This feature is triggered only in release branch, so the fix is not visible in the CI of this PR.
https://app.circleci.com/pipelines/github/pytorch/audio/9674/workflows/d0c9a6f1-8ca9-441a-a5f5-08926075fa39/jobs/553985?invite=true#step-104-193

The following outputs were observed when running it on local env.

* Before the change

```
$ BUILD_FFMPEG=0 BUILD_SOX=0 BUILD_CTC_DECODER=0 BUILD_RNNT=0 BUILD_KALDI=0 python setup.py clean bdist_wheel
```
```
-- Git branch: prototype-exclusion
-- Git SHA: 0af1edaa420c46be10292cbea7150c34ef80a0e1
-- Git tag: None
-- PyTorch dependency: torch
-- Building version 0.11.0+0af1eda
 --- Initializing submodules
 --- Initialized submodule
Excluding torchaudio.prototype from the package.
...
creating build/lib.macosx-11.0-arm64-3.9/torchaudio/prototype
creating build/lib.macosx-11.0-arm64-3.9/torchaudio/prototype/io
copying torchaudio/prototype/io/streamer.py -> build/lib.macosx-11.0-arm64-3.9/torchaudio/prototype/io
copying torchaudio/prototype/io/__init__.py -> build/lib.macosx-11.0-arm64-3.9/torchaudio/prototype/io
creating build/lib.macosx-11.0-arm64-3.9/torchaudio/prototype/pipelines
copying torchaudio/prototype/pipelines/__init__.py -> build/lib.macosx-11.0-arm64-3.9/torchaudio/prototype/pipelines
copying torchaudio/prototype/pipelines/rnnt_pipeline.py -> build/lib.macosx-11.0-arm64-3.9/torchaudio/prototype/pipelines
creating build/lib.macosx-11.0-arm64-3.9/torchaudio/prototype/ctc_decoder
copying torchaudio/prototype/ctc_decoder/ctc_decoder.py -> build/lib.macosx-11.0-arm64-3.9/torchaudio/prototype/ctc_decoder
copying torchaudio/prototype/ctc_decoder/__init__.py -> build/lib.macosx-11.0-arm64-3.9/torchaudio/prototype/ctc_decoder
warning: build_py: byte-compiling is disabled, skipping.
```

* After the change

```
$ BUILD_FFMPEG=0 BUILD_SOX=0 BUILD_CTC_DECODER=0 BUILD_RNNT=0 BUILD_KALDI=0 python setup.py clean bdist_wheel
```

```
-- Git branch: prototype-exclusion
-- Git SHA: 0af1edaa420c46be10292cbea7150c34ef80a0e1
-- Git tag: None
-- PyTorch dependency: torch
-- Building version 0.11.0+0af1eda
 --- Initializing submodules
 --- Initialized submodule
Excluding torchaudio.prototype from the package.
...
creating build/lib.macosx-11.0-arm64-3.9/torchaudio/models/wav2vec2
copying torchaudio/models/wav2vec2/__init__.py -> build/lib.macosx-11.0-arm64-3.9/torchaudio/models/wav2vec2
copying torchaudio/models/wav2vec2/model.py -> build/lib.macosx-11.0-arm64-3.9/torchaudio/models/wav2vec2
copying torchaudio/models/wav2vec2/components.py -> build/lib.macosx-11.0-arm64-3.9/torchaudio/models/wav2vec2
creating build/lib.macosx-11.0-arm64-3.9/torchaudio/models/wav2vec2/utils
copying torchaudio/models/wav2vec2/utils/__init__.py -> build/lib.macosx-11.0-arm64-3.9/torchaudio/models/wav2vec2/utils
copying torchaudio/models/wav2vec2/utils/import_huggingface.py -> build/lib.macosx-11.0-arm64-3.9/torchaudio/models/wav2vec2/utils
copying torchaudio/models/wav2vec2/utils/import_fairseq.py -> build/lib.macosx-11.0-arm64-3.9/torchaudio/models/wav2vec2/utils
warning: build_py: byte-compiling is disabled, skipping.
```

Pull Request resolved: https://github.com/pytorch/audio/pull/2225

Reviewed By: nateanl

Differential Revision: D34257128

Pulled By: mthrok

fbshipit-source-id: a3d6eca5803356e5aa3fe0eda82f6a9f5affb8e8
      a007e922
  3. 15 Feb, 2022 3 commits
    • moto's avatar
      Improve ffmpeg library discovery (#2204) · 963905e4
      moto authored 
      Summary:
This commit fixes the issue with ffmpeg discovery at build time.
The original implementation had issues like.

1. Wrong usage of FindFFMPEG, which caused mixture of ffmpeg libraries from system directory and user directory.
2. The optional `FFMPEG_ROOT` variable was not set within cmake.

The issue 1 is problematic when a user does not have a permission to
modify the environment. For example, an old version of ffmpeg, which is
installed in a directory managed by the system (such as `/usr/local/lib`),
then there is no way to specify a path in which user installs a supported version
of ffmpeg.

This commit changes the behavior by first searching the library
in `FFMPEG_ROOT` environment variables, then
resorting to the original behavior of searching the custom paths with
system default path.

Also this commirt removes support for `libavresample`, which is deprecated in
ffmpeg 4 and removed in ffmpeg 5.

Pull Request resolved: https://github.com/pytorch/audio/pull/2204

Reviewed By: carolineechen

Differential Revision: D34225769

Pulled By: mthrok

fbshipit-source-id: 95b0bfaaef31e2e69e6df29f789010f48a48210b
      963905e4
    • moto's avatar
      Update context building to not delay the inference (#2213) · 8e3c6144
      moto authored 
      Summary:
Updating the context cacher so that fetched audio chunk is used for inference immediately.

https://github.com/pytorch/audio/pull/2202#discussion_r802838174

Pull Request resolved: https://github.com/pytorch/audio/pull/2213

Reviewed By: hwangjeff

Differential Revision: D34235230

Pulled By: mthrok

fbshipit-source-id: 6e4aee7cca34ca81e40c0cb13497182f20f7f04e
      8e3c6144
    • hwangjeff's avatar
      Adjust Conformer args (#2223) · 411b5dcf
      hwangjeff authored 
      Summary:
Orders and names Conformer's initializer args to be more consistent with Emformer's.

Pull Request resolved: https://github.com/pytorch/audio/pull/2223

Reviewed By: mthrok

Differential Revision: D34226177

Pulled By: hwangjeff

fbshipit-source-id: 111c7ff27841aeac302ea5f6f7b50cc72c570829
      411b5dcf
  5. 11 Feb, 2022 7 commits
  7. 10 Feb, 2022 1 commit
  9. 09 Feb, 2022 2 commits
    • hwangjeff's avatar
      Clean up Emformer (#2207) · 87d7694d
      hwangjeff authored 
      Summary:
- Make `segment_length` a required argument rather than optional argument to force users to consciously choose input segment lengths for their use cases.
- Clarify expected input shapes in API documentation.
- Adjust `infer` tests to reflect expected usage.
- Add assertion for input shape for `infer`.

Pull Request resolved: https://github.com/pytorch/audio/pull/2207

Reviewed By: mthrok

Differential Revision: D34101205

Pulled By: hwangjeff

fbshipit-source-id: 1d1233d5edee5818d4669b4e47d44559e7ebb304
      87d7694d
    • hwangjeff's avatar
      Fix librosa calls (#2208) · e5d567c9
      hwangjeff authored 
      Summary:
Yesterday's release of librosa 0.9.0 made args keyword-only and changed default padding from "reflect" to "zero" for some functions. This PR adjusts callsites in our tutorials and tests accordingly.

Pull Request resolved: https://github.com/pytorch/audio/pull/2208

Reviewed By: mthrok

Differential Revision: D34099793

Pulled By: hwangjeff

fbshipit-source-id: 4e2642cdda8aae6d0a928befaf1bbb3873d229bc
      e5d567c9
  11. 04 Feb, 2022 1 commit
  13. 03 Feb, 2022 3 commits
  15. 02 Feb, 2022 5 commits
  17. 01 Feb, 2022 6 commits
  19. 31 Jan, 2022 1 commit
  21. 27 Jan, 2022 4 commits
    • hwangjeff's avatar
      Remove invalid token blanking logic from RNN-T decoder (#2180) · ed6256a2
      hwangjeff authored 
      Summary:
This PR removes logic in `RNNTBeamSearch` that blanks out joiner output values corresponding to special tokens, e.g. \<unk\>, \<eos\>, for the following reasons:
- Provided that the model was configured and trained properly, it shouldn't be necessary, e.g. the model would naturally produce low probabilities for special tokens if they don't exist in the training set.
- For our pre-trained LibriSpeech training pipeline, the removal of the logic doesn't affect evaluation WER on any of the dev/test splits.
- The existing logic doesn't generalize to arbitrary token vocabularies.
- Internally, it seems to have been acknowledged that this logic was introduced to compensate for quirks in other parts of the modeling infra.

Pull Request resolved: https://github.com/pytorch/audio/pull/2180

Reviewed By: carolineechen, mthrok

Differential Revision: D33822683

Pulled By: hwangjeff

fbshipit-source-id: e7047e294f71c732c77ae0c20fec60412f26f05a
      ed6256a2
    • Caroline Chen's avatar
      Add no lm support for CTC decoder (#2174) · 4c3fa875
      Caroline Chen authored 
      Summary:
Add support for CTC lexicon decoder without LM support by adding a non language model `ZeroLM` that returns score 0 for everything. Generalize the decoder class/API a bit to support this, adding it as an option for the kenlm decoder at the moment (will likely be separated out from kenlm when adding support for other kinds of LMs in the future)

Pull Request resolved: https://github.com/pytorch/audio/pull/2174

Reviewed By: hwangjeff, nateanl

Differential Revision: D33798674

Pulled By: carolineechen

fbshipit-source-id: ef8265f1d046011b143597b3b7c691566b08dcde
      4c3fa875
    • Zhaoheng Ni's avatar
      Refactor RNNT factory function to support num_symbols argument (#2178) · 2cb87c6b
      Zhaoheng Ni authored 
      Summary: Pull Request resolved: https://github.com/pytorch/audio/pull/2178

Reviewed By: mthrok

Differential Revision: D33797649

Pulled By: nateanl

fbshipit-source-id: 7a8f54294e7b5bd4d343c8e361e747bfd8b5b603
      2cb87c6b
    • moto's avatar
      Add `is_ffmpeg_available` in test (#2170) · 39fe9df6
      moto authored 
      Summary:
Part of https://github.com/pytorch/audio/issues/2164.
To make the tests introduced in https://github.com/pytorch/audio/issues/2164 skippable if ffmpeg features are not available,
this commit adds `is_ffmpeg_available`.

The availability of the features depend on two factors;
1. If it was enabled at build.
2. If the ffmpeg libraries are found at runtime.

A simple way (for OSS workflow) to detect these is simply checking if
`libtorchaudio_ffmpeg` presents and can be loaded without a failure.

To facilitate this, this commit changes the
`torchaudio._extension._load_lib` to return boolean result.

Pull Request resolved: https://github.com/pytorch/audio/pull/2170

Reviewed By: carolineechen

Differential Revision: D33797695

Pulled By: mthrok

fbshipit-source-id: 85e767fc06350b8f99de255bc965b8c92b8cfe97
      39fe9df6
  23. 26 Jan, 2022 1 commit