Fix Inconsistant input_feature length and attention_mask length in WhisperFeatureExtractor

[BakerBunker]

What does this PR do?

Fixes #39214

Before submitting

• This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
• Did you read the contributor guideline,
  Pull Request section?
• Was this discussed/approved via a Github issue or the forum? Please add a link
  to it if that's the case.
• Did you make sure to update the documentation with your changes? Here are the
  documentation guidelines, and
  here are tips on formatting docstrings.
• Did you write any new necessary tests?

Who can review?

@eustlb @zucchini-nlp

Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.

[vasqu]

Could you add a test? (just to make sure we don't have any regressions later on)

[zucchini-nlp]

run-slow: whisper

[github-actions]

This comment contains run-slow, running the specified jobs:

models: ['models/whisper']
quantizations: [] ...

[vasqu]

@BakerBunker Are you still interested in this PR?

[BakerBunker]

@BakerBunker Are you still interested in this PR?

yes, I'm a little busy these days.

[vasqu]

No worries, take your time

[BakerBunker]

@vasqu test case added

[vasqu]

Thanks, checking with models that might be affected then I'll merge

[vasqu]

run-slow: qwen2_5_omni,qwen2_audio,voxtral,whisper

[github-actions]

[For maintainers] Suggested jobs to run (before merge)

run-slow: whisper

[github-actions]

This comment contains run-slow, running the specified jobs:

models: ['models/qwen2_5_omni', 'models/qwen2_audio', 'models/voxtral', 'models/whisper']
quantizations: [] ...

[vasqu]

Seems like it does change the behavior of the models more than at least I expected. Can you add 🚨 to the title, as this is definitely a breaking change regardless.

Models:

• Voxtral has no new failures
• Same for qwen2 audio (although it just gets killed for whatever reason 👀)
• Whisper has one new test failing (looks like not a high diff tho)
• 2 new failures on qwen 2.5 omni cc @zucchini-nlp wdyt about the change?

[zucchini-nlp]

Yeah, oke for me as long as the tests in Qwen-Omni are updated before merging