ECAPATDNNOptions
Options & Configuration · AiDotNet.Audio.LanguageIdentification
Configuration options specific to ECAPA-TDNN language identification.
Inheritance: ModelOptions → LanguageIdentifierOptions → ECAPATDNNOptions
Known derived types: VoxLingua107Options
Constructors
| Constructor | Summary |
|---|---|
ECAPATDNNOptions() |
Properties
| Property | Summary |
|---|---|
bool UseChannelAttention { get; set; } |
Gets or sets whether to use channel attention. |
int AttentionChannels { get; set; } |
Gets or sets the bottleneck width of the attentive statistics pooling. |
int AttentionHeads { get; set; } |
Gets or sets the number of attention heads. |
int Res2NetScale { get; set; } |
Gets or sets the Res2Net scale: the number of channel splits in each SE-Res2Block. |
int SeChannels { get; set; } |
Gets or sets the squeeze-excitation bottleneck width. |
int TdnnChannels { get; set; } |
Gets or sets the number of TDNN channels. |
int[] Dilations { get; set; } |
Gets or sets the dilation of each ECAPA-TDNN stage. |
int[] KernelSizes { get; set; } |
Gets or sets the kernel width of each ECAPA-TDNN stage, aligned with Dilations. |