v0.43.1

fix: ensemble separation ignores custom_output_names, misclassifies s…

v0.43.0

chore: bump version to 0.43.0 (#274)

* chore: bump version to 0.43.0

New features in this release:
- Cloud Run GPU deployment support (deploy_cloudrun.py, Dockerfile.cloudrun)
- Ensemble preset support in API client (-p/--preset flag in CLI)
- Preset parameter in separate_audio() and separate_audio_and_wait()
- Stem marker stripping for chained separations

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* Trigger CI: runners restarted, idle check paused

---------

Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

v0.42.1

chore: bump version to 0.42.1 (#271)

Patch release for:
- Fix chunk overwrite when using custom_output_names with chunk_duration (#259, #270)
- Clarify demucs segment_size "Default" documentation (#264, #270)

Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

v0.42.0

feat: multi-model ensemble separation with 9 community-curated preset…

v0.41.1

fix: RFC 5987 encoding for Unicode filenames in Content-Disposition h…

v0.41.0

Update pyproject.toml

v0.40.0

Preserve bit depth of input audio (don't down sample everything to 16…

v0.39.1

Fix multi-stem MDXC bug (#242)

v0.39.0

Updated reference spectrogram to reflect fixed MDX implementation (#238)

v0.38.1

Fix mel regression (#235)

* Fixed MelBandRoformer modal loading regression

* Added mel_band_roformer_karaoke_aufr33_viperx_sdr_10.1956.ckpt to integration test