AcousticConditioning<T>

Models & Types · AiDotNet.TextToSpeech

The phoneme hidden sequence after a variance-adaptor model adds its acoustic conditions, with what its decoder is conditioned on and any auxiliary loss the conditioning contributes in training.

Implements: IEquatable<AcousticConditioning<T>>

Constructors

Constructor Summary
AcousticConditioning(Tensor<T> Hidden, Tensor<T> DecoderCondition, Tensor<T> AuxiliaryLoss) The phoneme hidden sequence after a variance-adaptor model adds its acoustic conditions, with what its decoder is conditioned on and any auxiliary loss the conditioning contributes in training.

Properties

Property Summary
Tensor<T> AuxiliaryLoss { get; set; } A scalar loss term added to the objective (AdaSpeech’s phoneme-level predictor loss); null for none.
Tensor<T> DecoderCondition { get; set; } The vector conditional layer normalizations in the decoder read (AdaSpeech’s speaker embedding); null for an unconditioned decoder.
Tensor<T> Hidden { get; set; } The conditioned phoneme hidden sequence the variance adaptor reads.

Methods

Method Summary
AcousticConditioning<T> <Clone>$()
bool Equals(AcousticConditioning<T> other)
bool Equals(object obj)
bool op_Equality(AcousticConditioning<T> left, AcousticConditioning<T> right)
bool op_Inequality(AcousticConditioning<T> left, AcousticConditioning<T> right)
int GetHashCode()
string ToString()
void Deconstruct(Tensor<T> Hidden, Tensor<T> DecoderCondition, Tensor<T> AuxiliaryLoss)