DDSP
Learn how to control oscillators, noise and filters from recordings for resynthesis and timbre transfer.
Neural instrumentsOpen-source code
- Start here
- Start with a clean single-note instrument recording. Listen for pitch tracking errors before changing the synthesis model.
- Training data
- Depends on the experiment or checkpoint; the framework supports preparing your own instrument recordings.
- Training rights
- Framework access does not license example audio or every pretrained model. A custom recording collection needs documented permissions.
- License & output terms
- Code: Apache 2.0; inspect the selected checkpoint separately.
- Limitations
- Classic instrument examples assume pitched, often monophonic material. Pitch extraction and synthesis assumptions shape the result.
- Technique
- Differentiable signal processing
- Runtime
- Local training and inference; deployment requirements depend on the chosen architecture.
- Parameters
- Not itemized for this release/family. Parameter sizes shown elsewhere may exclude encoders, decoders or other runtime components.
Sources reviewed . Training and rights notes summarize provider disclosures. A model license does not clear every source recording or generated output. Parameter counts refer to the named variant or component, not total runtime memory.