Releases: nomadkaraoke/python-audio-separator
Releases · nomadkaraoke/python-audio-separator
v0.33.0
v0.32.0
Added waveform & spectrogram verification to integration tests (#209) * Added waveform & spectrogram verification to integration tests * Raised minimum similarity threshold for waveform/spectrogram image comparison in integration tests, except for demucs which is pretty inconsistent with itself * Updated dependencies, moved to using SSIM for wave/spect comparison in integration test * Attempt to fix integration test volatility with demucs by fixing Y axis for waveform images, but spectrograms are still all over the place * Added integration test verifying the output dir and custom output names work as expected * Split waveform and spec thresholds, lowered all, added images to test artifacts for CI * Split DEFAULT_SIMILARITY_THRESHOLD to override spectrogram threshold * Only run integration tests in integration workflow...
v0.31.3
Added integration test covering real execution with all model archite…
v0.31.2
Update pyproject.toml
v0.31.1
Bump version for release
v0.31.0
Update to weekly onnx and numpy 2.x (#200) * Committing to branch to check tests cross platform * Fixed lockfile` * Attempt to prevent onnx 1.17 install on python 3.13 despite transitive dep * Test whether python version excudes onnx transitive * Switched to using onnx2torch-py313 instead of onnx2torch to resolve python 3.13 dependency issues
v0.30.2
Maybe fix the seek during model hash on windows (#198) * Maybe fix the seek during model hash on windows * Fixed hash calculation bug
v0.30.1
Added AUDIO_SEPARATOR_MODEL_DIR env var to override model dir. Relaxe…
v0.30.0
Update pyproject.toml
v0.29.0
Added new models from PR