DemucsNoiseOptions

Options & Configuration · AiDotNet.Audio.Effects

Configuration options for the Demucs for Noise model.

Inheritance: ModelOptions → DemucsNoiseOptions

For Beginners

Demucs for Noise is like using a music separator to pull apart “speech” and “noise” tracks from a recording. The original Demucs separates vocals, drums, bass, and other instruments. This version separates clean speech from background noise - perfect for cleaning up phone calls, podcasts, and video meetings.

How It Works

Demucs for Noise (Defossez et al., 2020, Meta) adapts the Demucs architecture (originally for music source separation) for real-time noise suppression. It operates in the time domain with a U-Net encoder-decoder structure and skip connections, achieving high-quality noise removal at low latency (40ms).

Constructors

Constructor Summary
DemucsNoiseOptions()

Properties

Property Summary
OnnxModelOptions OnnxOptions { get; set; } Gets or sets the ONNX runtime options.
double DropoutRate { get; set; } Gets or sets the dropout rate.
double LearningRate { get; set; } Gets or sets the learning rate.
int ChannelGrowth { get; set; } Gets or sets the channel growth factor per depth level.
int Depth { get; set; } Gets or sets the encoder/decoder depth (number of layers).
int HiddenChannels { get; set; } Gets or sets the initial hidden channels.
int KernelSize { get; set; } Gets or sets the kernel size for the encoder convolutions.
int LSTMHiddenSize { get; set; } Gets or sets the LSTM hidden size for the bottleneck.
int NumChannels { get; set; } Gets or sets the number of audio channels.
int NumLSTMLayers { get; set; } Gets or sets the number of LSTM layers in the bottleneck.
int SampleRate { get; set; } Gets or sets the audio sample rate in Hz.
int Stride { get; set; } Gets or sets the stride for the encoder convolutions.
string ModelPath { get; set; } Gets or sets the path to the ONNX model file.
string Variant { get; set; } Gets or sets the model variant (“small”, “base”, “large”).