PANNs
Detect sound-event classes or extract general audio embeddings, including music and instruments.
UnderstandingOpen-source code
- Start here
- Use it for broad sound-event labels. For detailed chord, key or note questions, choose a task designed for that representation.
- Training data
- AudioSet's labeled YouTube sound segments.
- Training rights
- AudioSet annotations do not transfer ownership of the underlying YouTube recordings; available audio changes over time.
- License & output terms
- Code: MIT. Check the linked checkpoint release for weight terms.
- Limitations
- Broad sound-event classes do not describe detailed harmony, form or note content; labels can be noisy.
- Technique
- Pretrained audio convolutional networks
- Runtime
- Local PyTorch inference; architecture choice trades memory and speed for accuracy.
- 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.