🚨🚧 FeatureExtractor → AudioProcessor - #44394
Draft
eustlb wants to merge 70 commits into
Draft
Conversation
|
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update. |
eustlb
changed the base branch from
run_update_tiny_002
to
refactor-improc-backends
March 2, 2026 17:48
2 tasks
6 tasks
This was referenced Apr 29, 2026
…ractionMixin; ruff format
…ackend-dicts
- format_args_docstring: unwrap the {feature_extractor_class} backend-dict (torch default),
not just {image_processor_class} (regression from merged image-backend refactor #43514)
- MODALITY_TO_BASE_CLASS_MAPPING: accept Torch/NumpyAudioBackend for the feature_extractor
and audio_processor slots so new XxxAudioProcessor classes pass ProcessorMixin validation
Add SpectrogramConfig.preemphasis_mode ('per_frame' default | 'waveform'). In 'waveform'
mode the base _stft applies preemphasis to the raw waveform (first sample unchanged) and
zeroes padded samples via audio_ranges, and preemphasis no longer forces manual framing.
Parakeet, Parakeet-numpy and Nemotron (inherits Parakeet) drop their _stft and
_needs_manual_framing overrides entirely; Cohere's _stft shrinks to just its dither.
Verified bit-identical output on all affected models; per-frame preemphasis models
(AST, seamless_m4t) unchanged. (ADR-0004 rule-of-two: pattern shared by 3 models.)
…gram Superseded by audio_utils.py + audio_processing_backends.py (the 'temporarily use separate backends files' iteration). No references anywhere in src/tests/docs/benchmarks.
4 tasks
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What does this PR do?
Models that have a
feature_extractor_xxxand that should be addressed by this PR:should cover
TODO