VarianceTargets<T>
Layers · AiDotNet.NeuralNetworks.Layers
Ground-truth variance for one utterance (FastSpeech 2 teacher forcing).
Constructors
| Constructor | Summary |
|---|---|
VarianceTargets() |
Properties
| Property | Summary |
|---|---|
Tensor<T> Energy { get; set; } |
Energy per frame. |
Tensor<T> Pitch { get; set; } |
F0 per frame in Hz, with unvoiced frames filled (continuous contour). |
int[] Durations { get; set; } |
Frames per token. |