ECAPATDNNSpeakerOptions
Options & Configuration · AiDotNet.Audio.Speaker
Configuration options for the ECAPA-TDNN speaker verification and embedding model.
Inheritance: ModelOptions → ECAPATDNNSpeakerOptions
For Beginners
ECAPA-TDNN creates a “voiceprint” for any speaker. It processes audio through special layers that focus on the most important voice characteristics. The result is a compact vector (embedding) that uniquely identifies a speaker’s voice.
How It Works
ECAPA-TDNN (Desplanques et al., Interspeech 2020) is a state-of-the-art speaker embedding model that extends x-vector architecture with Squeeze-Excitation blocks, multi-layer feature aggregation, and channel- and context-dependent statistics pooling. Achieves 0.87% EER on VoxCeleb1 test set.
Constructors
| Constructor | Summary |
|---|---|
ECAPATDNNSpeakerOptions() |
Properties
| Property | Summary |
|---|---|
OnnxModelOptions OnnxOptions { get; set; } |
Gets or sets the ONNX runtime options. |
double DefaultThreshold { get; set; } |
Gets or sets the default cosine similarity threshold for verification. |
double DropoutRate { get; set; } |
Not used: ECAPA-TDNN (Desplanques et al. |
double LearningRate { get; set; } |
Gets or sets the learning rate for training. |
double MinDurationSeconds { get; set; } |
Gets or sets the minimum audio duration in seconds for reliable embedding. |
double WeightDecay { get; set; } |
Gets or sets the weight decay for regularization. |
int AttentionChannels { get; set; } |
Gets or sets the bottleneck width of the attentive statistics pooling. |
int EmbeddingDim { get; set; } |
Gets or sets the output embedding dimension. |
int FftSize { get; set; } |
Gets or sets the FFT window size in samples. |
int HopLength { get; set; } |
Gets or sets the hop length between frames in samples. |
int NumMels { get; set; } |
Gets or sets the number of mel filterbank channels. |
int PoolingDim { get; set; } |
The width the attentive statistics pooling reads: the MFA convolution’s, the last entry of Channels (1536 in the paper). |
int Res2NetScale { get; set; } |
Gets or sets the Res2Net scale factor. |
int SEBottleneckDim { get; set; } |
Gets or sets the SE (Squeeze-Excitation) bottleneck dimension. |
int SampleRate { get; set; } |
Gets or sets the expected audio sample rate in Hz. |
int[] Channels { get; set; } |
Gets or sets the channel dimensions for each TDNN block. |
int[] Dilations { get; set; } |
Gets or sets the dilation factors for each TDNN block. |
int[] KernelSizes { get; set; } |
Gets or sets the TDNN kernel sizes (dilation factors). |
string ModelPath { get; set; } |
Gets or sets the path to the ONNX model file. |