MPSENet<T>

Models & Types · AiDotNet.Audio.Enhancement

MP-SENet (Multi-Path Speech Enhancement Network) model (Lu et al., 2023).

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

Implements: IAudioEnhancer<T>, IGeneratedParameterRegistrar<T>

For Beginners

Sound has two components: loudness (magnitude) and timing (phase). Most enhancers only fix loudness, leaving timing artifacts. MP-SENet fixes both at once using two parallel paths that communicate, producing cleaner, more natural audio.

Usage:

How It Works

MP-SENet predicts both magnitude and phase of the complex spectrogram using parallel estimation paths with cross-domain fusion. It achieves PESQ 3.60 on VoiceBank+DEMAND, surpassing prior single-channel speech enhancement methods.

Constructors

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

Properties

Property Summary
double EnhancementStrength { get; set; }
int LatencySamples { get; }
int NumChannels { get; }

Methods

Method Summary
ModelMetadata<T> GetModelMetadata()
Task<Tensor<T>> EnhanceAsync(Tensor<T> noisyAudio, CancellationToken cancellationToken)
Tensor<T> Enhance(Tensor<T> noisyAudio)
Tensor<T> EnhanceWithReference(Tensor<T> audio, Tensor<T> reference)
Tensor<T> ProcessChunk(Tensor<T> audioChunk)
void EstimateNoiseProfile(Tensor<T> noiseOnlyAudio)