Skip to content

fix issue with ctc_forward_step #2505

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Merged
mravanelli merged 5 commits into from
Apr 13, 2024
Merged

fix issue with ctc_forward_step #2505

mravanelli merged 5 commits into from
Apr 13, 2024

Conversation

@Adel-Moumen
Copy link
Collaborator

@Adel-Moumen Adel-Moumen commented Apr 11, 2024 •
edited
Loading

What does this PR do?

This PR aims to fix one issue with the ctc_segmentation module due to the introduction of the new beam search part. Indeed, there's no longer the function ctc_forward_step inside the beam search. To address this issue, I added new code to make it work with SB 1.0.

Fixes #2440

repro:

from speechbrain.inference.ASR import EncoderDecoderASR
from speechbrain.alignment.ctc_segmentation import CTCSegmentation
# load an ASR model
pre_trained = "speechbrain/asr-transformer-transformerlm-librispeech"
asr_model = EncoderDecoderASR.from_hparams(source=pre_trained)
aligner = CTCSegmentation(asr_model, kaldi_style_text=False)
# load data
audio_path = "/content/speechbrain/tests/samples/single-mic/example1.wav"
text = ["THE BIRCH CANOE", "SLID ON THE", "SMOOTH PLANKS"]
segments = aligner(audio_path, text, name="example1")

Before

---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
[<ipython-input-4-cb3639878d03>](https://localhost:8080/#) in <cell line: 6>()
      4 pre_trained = "speechbrain/asr-transformer-transformerlm-librispeech"
      5 asr_model = EncoderDecoderASR.from_hparams(source=pre_trained)
----> 6 aligner = CTCSegmentation(asr_model, kaldi_style_text=False)
      7 # load data
      8 audio_path = "/content/speechbrain/tests/samples/single-mic/example1.wav"

1 frames
[/usr/local/lib/python3.10/dist-packages/torch/nn/modules/module.py](https://localhost:8080/#) in __getattr__(self, name)
   1693             if name in modules:
   1694                 return modules[name]
-> 1695         raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
   1696 
   1697     def __setattr__(self, name: str, value: Union[Tensor, 'Module']) -> None:

AttributeError: 'S2STransformerBeamSearcher' object has no attribute 'ctc_forward_step'

After

The dictionary has 5000 tokens with a max length of 15. This may lead to low alignment performance and low accuracy.
example1_0000 example1 0.78 1.61 -5.7213 THE BIRCH CANOE
example1_0001 example1 1.61 1.85 -11.5366 SLID ON THE
example1_0002 example1 1.85 2.81 -3.2060 SMOOTH PLANKS
Before submitting
  • Did you read the contributor guideline?
  • Did you make sure your PR does only one thing, instead of bundling different changes together?
  • Did you make sure to update the documentation with your changes? (if necessary)
  • Did you write any new necessary tests? (not for typos and docs)
  • Did you verify new and existing tests pass locally with your changes?
  • Did you list all the breaking changes introduced by this pull request?
  • Does your code adhere to project-specific code style and conventions?

PR review

Reviewer checklist
  • Is this pull request ready for review? (if not, please submit in draft mode)
  • Check that all items from Before submitting are resolved
  • Make sure the title is self-explanatory and the description concisely explains the PR
  • Add labels and milestones (and optionally projects) to the PR so it can be classified
  • Confirm that the changes adhere to compatibility requirements (e.g., Python version, platform)
  • Review the self-review checklist to ensure the code is ready for review

@Adel-Moumen Adel-Moumen marked this pull request as draft April 11, 2024 19:59
@Adel-Moumen Adel-Moumen marked this pull request as ready for review April 11, 2024 20:32
@mravanelli mravanelli self-requested a review April 12, 2024 01:32
@mravanelli mravanelli added the bug Something isn't working label Apr 12, 2024
@mravanelli
Copy link
Collaborator

mravanelli commented Apr 13, 2024

Thank you @Adel-Moumen, I tested the latest version and everything seems fixed!

@mravanelli mravanelli merged commit d187cba into speechbrain:develop Apr 13, 2024
@Mickaelh51
Copy link

Mickaelh51 commented Apr 13, 2024

Hi,
Thanks a lot for this PR.
It works like a charm
++

pchampio pushed a commit to pchampio/speechbrain that referenced this pull request Apr 17, 2024
@Adel-Moumen @pchampio
fix issue with ctc_forward_step (speechbrain#2505)
4554c1e 
* fix issue with ctc_forward_step

* improve documentation

* add check

* add checks

* fix checks
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@mravanelli mravanelli mravanelli approved these changes

Assignees

@Adel-Moumen Adel-Moumen

Labels

bug Something isn't working

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

'S2STransformerBeamSearcher' object has no attribute 'ctc_forward_step'

3 participants

@Adel-Moumen @mravanelli @Mickaelh51