Expand description
Source separation (classical masking now; learned = NeedsWeights). Re-exports only (AU-GEN).
apply_soft_mask / binary_mask_from_ratio are real classical spectral masking and work
now over caller-owned buffers. separate_learned is a fail-closed stub for demucs-class
learned separation that always returns crate::types::AudioError::BackendUnavailable — it
never fabricates stems.
Re-exports§
pub use learned::separate_learned;pub use mask::apply_soft_mask;pub use mask::binary_mask_from_ratio;
Modules§
- learned
- Learned (demucs / open-unmix / band-split RNN class) source separation — fail-closed stub.
- mask
- Classical spectral masking for source separation. Given a mixture magnitude spectrum and a
per-bin soft mask (each element in
[0, 1]), the separated stem magnitude is the element-wise productstem = mixture ⊙ mask. This is the algorithmic core of ideal-ratio / Wiener-style masking; it invents nothing — it only re-weights energy the mixture already contains. Learned mask estimation (a demucs/openunmix network) is a separate,NeedsWeightsconcern (seelearned.rs).