silverdaw/htdemucs-ft-onnx
Audio-to-Audio • Updated
Silverdaw
non-profit
AI & ML interests
None defined yet.
Recent Activity
Organization Card
Silverdaw
Open-source stem-separation models for Silverdaw — a Windows DAW for bedroom DJs, producers, and mixers making remixes and mashups.
These repositories host the ONNX models Silverdaw downloads on first use. They are re-hosted here (with their original MIT licences and full attribution) so the app can fetch every model it needs from one stable, self-owned location.
Models
| Repository | Role | Stems |
|---|---|---|
mel-band-roformer-vocals-onnx |
Primary vocal model | vocals |
bs-roformer-rhythm-onnx |
Primary drums + bass model | drums, bass |
htdemucs-ft-onnx |
Backup 4-stem model | vocals, drums, bass, other |
All models are MIT-licensed and run offline (non-real-time) on CPU or on DirectX 12 GPUs via ONNX Runtime's DirectML execution provider. Each repository's model card documents the I/O contract, provenance, and upstream attribution.
Links
- App & source: https://github.com/irarainey/silverdaw
datasets 0
None public yet