Separator v0.1.2

Separator 0.1.2

Native audio separation for Windows x64, Linux x64 and Apple Silicon macOS 14+.

• Instrumental Ultra: Inst v1e+ and Becruily using upstream max-spec ensembling.
• Vocal Ultra: Resurrection and Big Beta6x using upstream FFT averaging.
• Select individual models or create weighted custom ensembles with advanced inference controls.
• Persistent jobs/presets, cancellation/retry, lossless exports, synchronized playback and blind range comparisons.
• Private CPU/MPS runtime and optional verified NVIDIA CUDA installation; no system Python changes.
• Source-built FFmpeg, original third-party notices and exact corresponding codec sources included.

Choose the .exe installer on Windows, .AppImage or .deb on Linux, and .dmg on Apple Silicon. .app.tar.gz and .sig files serve the signed in-app updater. SHA256SUMS covers every published installer/update/source asset.

The RTX 3070 8 GB development system executed both Ultra ensembles in float32, individual RoFormer and Demucs models, native playback/comparison and cancellation, and isolated CPU-container processing. These synthetic execution fixtures do not establish a universal perceptual model ranking. Native CI builds and checks each installed private runtime and its audio codecs; Windows CUDA and Apple MPS inference still require matching hardware validation.

Updater signatures are verified against the committed public key. Apple Developer ID/notarization and Windows Authenticode certificates were not supplied for this release; the installers do not claim these OS signatures. Model weights download on request and retain their authors' licensing terms.