IAudioEnhancer<T>
Interfaces · AiDotNet.Interfaces
Defines the contract for audio enhancement models that improve audio quality.
Known implementations: AudioEnhancerBase<T>, AudioSuperResolution<T>, BandSplitRNNEnhancer<T>, CMGAN<T>, ConvTasNet<T>, DCCRN<T>, DeepFilterNet<T>, DemucsNoise<T>, FRCRN<T>, FullSubNetPlus<T>, MPSENet<T>, NeuralNoiseReducer<T>, NeuralParametricEQ<T>, RoomImpulseResponse<T>, SpikingFullSubNet<T>, TFGridNet<T>
For Beginners
Audio enhancement is like photo editing for sound!
Common use cases:
- Cleaning up podcast recordings (removing AC hum, keyboard clicks)
- Improving phone call quality (reducing background noise)
- Restoring old recordings (removing tape hiss, crackle)
- Video conferencing (echo cancellation, noise suppression)
- Hearing aids (speech enhancement in noisy environments)
How it works (simplified):
- Analyze the audio to identify “noise” vs “signal”
- Create a filter that reduces noise while keeping the signal
- Apply the filter to produce cleaner audio
Modern approaches use neural networks that learn what clean audio should sound like, producing much better results than traditional methods.
How It Works
Audio enhancement encompasses various techniques to improve audio quality:
- Noise Reduction: Remove background noise while preserving speech/music
- Speech Enhancement: Improve speech intelligibility and quality
- Dereverberation: Remove room echo and reverb artifacts
- Echo Cancellation: Remove acoustic echo in communication systems
- Bandwidth Extension: Extend frequency range of narrowband audio
Properties
| Property | Summary |
|---|---|
double EnhancementStrength { get; set; } |
Gets or sets the enhancement strength (0.0 = no enhancement, 1.0 = maximum). |
int LatencySamples { get; } |
Gets the processing latency in samples. |
int NumChannels { get; } |
Gets the number of audio channels supported. |
int SampleRate { get; } |
Gets the sample rate this enhancer operates at. |
Methods
| Method | Summary |
|---|---|
Tensor<T> Enhance(Tensor<T> audio) |
Enhances audio quality by reducing noise and artifacts. |
Tensor<T> EnhanceWithReference(Tensor<T> audio, Tensor<T> reference) |
Enhances audio with a reference signal for echo cancellation. |
Tensor<T> ProcessChunk(Tensor<T> audioChunk) |
Processes audio in real-time streaming mode. |
void EstimateNoiseProfile(Tensor<T> noiseOnlyAudio) |
Estimates the noise profile from a segment of audio. |
void ResetState() |
Resets internal state for streaming mode. |