BandSplitRNN<T>

Models & Types · AiDotNet.Audio.SourceSeparation

BandSplitRNN for music source separation (Luo and Yu, 2023).

Inheritance: NeuralNetworkBase<T> → AudioNeuralNetworkBase<T> → BandSplitRNN<T>

Implements: IMusicSourceSeparator<T>

For Beginners

BandSplitRNN separates a mixed song into individual instruments (vocals, drums, bass, other) by dividing audio frequencies into bands—like splitting an equalizer into sections. Each section is cleaned up by one specialist RNN, then a second RNN checks that the sections work together consistently. The result is clean, separated audio for each instrument.

Usage:

How It Works

BandSplitRNN splits the spectrogram into non-overlapping frequency bands, processes each band independently with a shared band-level RNN, applies cross-band fusion via a sequence-level RNN, and estimates source-specific masks. This dual-RNN design achieves 10.0+ dB SDR on MUSDB18-HQ, establishing a strong RNN-based baseline for music source separation.

Constructors

Constructor Summary
BandSplitRNN(NeuralNetworkArchitecture<T> architecture, BandSplitRNNOptions options, IGradientBasedOptimizer<T, Tensor<T>, Tensor<T>> optimizer)
BandSplitRNN(NeuralNetworkArchitecture<T> architecture, string modelPath, BandSplitRNNOptions options)

Properties

Property Summary
IReadOnlyList<string> SupportedSources { get; }
int NumStems { get; }

Methods

Method Summary
ModelMetadata<T> GetModelMetadata()
SourceSeparationResult<T> Separate(Tensor<T> audio)
Task<SourceSeparationResult<T>> SeparateAsync(Tensor<T> audio, CancellationToken ct)
Tensor<T> ExtractSource(Tensor<T> audio, string source)
Tensor<T> GetSourceMask(Tensor<T> audio, string source)
Tensor<T> Remix(SourceSeparationResult<T> separationResult, IReadOnlyDictionary<string, double> sourceVolumes)
Tensor<T> RemoveSource(Tensor<T> audio, string source)