Pulling remote to local#1
Open
vizsatiz wants to merge 129 commits into
Open
Conversation
support timestamp for numbers.
chore: remove deprecated VAD_SEGMENTATION_URL
prevent errors
chore: handle empty segments_list case in silero
chore: fix variable naming inconsistency from `segments` to `segments_list`
Add models_cache_only param
…ation and publishing
[New File] benchmark testing
Batch wrap
Revert "Batch wrap"
Pass through the average log probability (transcription confidence score) from ctranslate2 to the final segment output. The field is NotRequired so existing code constructing segments without it remains valid. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
PR #986 ("support timestamps for numbers") introduced three changes that together broke CTC forced alignment: 1. Unknown chars (numbers, punctuation) were replaced with '*' wildcards mapped to token -1. get_wildcard_emission() scored these using torch.max() over all non-blank emissions, so wildcards greedily matched any speech-like signal in the segment window. 2. get_trellis() was rewritten with a different shape (num_frame, num_tokens) and incompatible initialization, discarding the original SoS-offset design from the PyTorch forced alignment tutorial. 3. backtrack() was replaced with backtrack_beam(), which always starts backtracking from the last frame of the segment window. The original backtrack() used torch.argmax() on the last token column to determine the starting frame. With padded segment boundaries (silence before/after speech), the new implementation spread all tokens across the full window, placing the first word at the start of the silence instead of the speech. This commit restores the original PyTorch tutorial implementation: - Unknown chars are skipped; words with only unknown chars become unalignable and get no timestamps (handled by interpolate_nans) - get_trellis: restored (num_frame+1, num_tokens+1) shape with SoS offset - backtrack: restored torch.argmax-based starting frame - Removed backtrack_beam, get_wildcard_emission, BeamState, Path Verified: v3.3.0 (pre-#986) produced correct timestamps with padded segment boundaries; this fix reproduces that behavior. Fixes #1220 Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
…ktrack The original code accepted blank_id as a parameter but used hardcoded 0 in two places, breaking alignment for HuggingFace models where the blank token is [pad] (not index 0). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Add optional `progress_callback: Callable[[float], None]` parameter to the three public API functions for real-time progress tracking. Each callback receives 0-100% for its own stage independently. Diarization wraps the callback into pyannote's internal hook protocol, keeping pyannote internals fully encapsulated. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Characters not in the alignment model's vocabulary (digits, symbols, foreign script) were excluded from CTC alignment and got no timestamps. Fix: extend the emission matrix with a wildcard column (max non-blank score per frame) so unknown chars participate in alignment via the existing get_trellis/backtrack — no changes to the CTC internals. Interpolation fallback fills any remaining gaps as a safety net. Unlike PR #986, this does not rewrite get_trellis or backtrack, so it avoids the regression that caused #1220. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Reproduces the exact reported bug: '4,9' embedded in text gets no start/end/score because digits and commas are not in the alignment model dictionary. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Add pytest to optional dev dependencies and create a dedicated tests.yml workflow to run pytest across Python 3.10-3.13. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
On environments like Google Colab where torchvision is pre-installed, `pip install whisperx` upgrades torch to 2.8.0 but leaves the old incompatible torchvision behind. The version mismatch causes `RuntimeError: operator torchvision::nms does not exist` which cascades through transformers' import chain into the confusing `ModuleNotFoundError: Could not import module 'Wav2Vec2ForCTC'`. Fixes #1339 Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
torchcodec 0.10.0 requires torch 2.10; only 0.6.x-0.7.x are compatible with torch 2.8. Without the cap, pip resolves to 0.10.0+ causing runtime errors. Platform markers are preserved since torchcodec has no aarch64 Linux wheels for these versions. Fixes #1392 Co-Authored-By: ssakkout <145016316+ssakkout@users.noreply.github.com> Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
`--interpolate_method ignore` crashed with ValueError because pandas doesn't support "ignore" as a Series.interpolate() method. Return the series unchanged so NaN values are preserved — callers already guard with pd.notna() before assigning timestamps. Closes #1368 Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Bumps [nltk](https://github.com/nltk/nltk) from 3.9.2 to 3.9.4. - [Changelog](https://github.com/nltk/nltk/blob/develop/ChangeLog) - [Commits](nltk/nltk@3.9.2...3.9.4) --- updated-dependencies: - dependency-name: nltk dependency-version: 3.9.4 dependency-type: direct:production ... Signed-off-by: dependabot[bot] <support@github.com>
Add GitHub Actions security audit via zizmor. Fix all existing workflow findings: pin actions to commit hashes, add persist-credentials: false, set explicit least-privilege permissions, disable uv cache in release workflow to prevent cache poisoning. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Remove third-party action in favor of pre-installed gh CLI. Pass release tag via env var to prevent template injection. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
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
20 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.
No description provided.