Pin torchaudio below 2.9 and document global install from source - #38
Merged
Conversation
pyannote-audio still calls torchaudio.list_audio_backends(), which was removed in torchaudio 2.9, so fresh installs crash during transcription with AttributeError. Cap torchaudio to <2.9 until this is fixed upstream (pyannote/pyannote-audio#1974). Also cap requires-python to <3.14: torchaudio 2.8 has no Python 3.14 wheels, so uvx would otherwise pick 3.14 and fail to resolve. Matches the classifiers, which already only list 3.12 and 3.13. Fixes #34
After 'uv sync' the binary only exists inside .venv/bin and is not on PATH. Document 'uv tool install --editable .' as the follow-up step. Fixes #26
A brew formula isn't practical for the PyTorch/llama-cpp-python dependency stack, so document the uv-based alternative instead. Refs #33
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
torchaudio>=2.8,<2.9: pyannote-audio still callstorchaudio.list_audio_backends(), which was removed in torchaudio 2.9, so every fresh install (e.g.uvx ownscribe) crashes during transcription. The cap is temporary until Is torchaudio 2.9 supported? pyannote/pyannote-audio#1974 is fixed upstream.requires-pythonto<3.14: torchaudio 2.8 has no Python 3.14 wheels, so without the cap uvx picks 3.14 and resolution fails. Matches the existing classifiers (3.12/3.13 only).uv tool install --editable .step to the "From source" section so theownscribecommand lands on PATH.brew install uv && uv tool install 'ownscribe[all]'since a brew formula isn't practical for this dependency stack.The uv.lock churn is only dropped Python 3.14 marker branches — still the same 144 packages, torchaudio stays at 2.8.0.
Fixes #34
Fixes #26
Refs #33
Test plan
uv lockresolves cleanly, torchaudio pinned at 2.8.0uv run pytest: 243 passed🤖 Generated with Claude Code