FastSpeech2Options

Options & Configuration · AiDotNet.TextToSpeech.Classic

Options for FastSpeech 2 (variance adaptor with pitch, energy, and duration predictors).

Inheritance: ModelOptions → TtsModelOptions → AcousticModelOptions → FastSpeech2Options

Known derived types: AdaSpeechOptions, PromptTTSOptions

For Beginners

These options configure the FastSpeech2 model. Default values follow the original paper settings.

How It Works

Defaults are the paper’s LJSpeech configuration (Ren et al. 2021, App. A, Table 7).

Constructors

Constructor Summary
FastSpeech2Options() Initializes a new instance with default values.
FastSpeech2Options(FastSpeech2Options other) Initializes a new instance by copying from another instance.

Properties

Property Summary
bool UseEnergyPredictor { get; set; } Gets or sets whether to use energy prediction (the paper’s ablation removes it).
bool UsePitchPredictor { get; set; } Gets or sets whether to use pitch prediction (the paper’s ablation removes it).
double PitchMaxHz { get; set; } Gets or sets the highest pitch bin edge in Hz (WORLD DIO’s F0 ceiling, 800 Hz).
double PitchMinHz { get; set; } Gets or sets the lowest pitch bin edge in Hz (WORLD DIO’s F0 floor, 71 Hz).
double VariancePredictorDropout { get; set; } Gets or sets the variance predictor dropout (0.5).
int DecoderDim { get; set; } Decoder output width in mel channels; an alias for MelChannels.
int FftFilterSize { get; set; } Gets or sets the filter size of the FFT blocks’ first convolution (1024).
int NumEnergyBins { get; set; } Gets or sets the number of energy bins for quantization (256, uniform).
int NumPitchBins { get; set; } Gets or sets the number of pitch bins for quantization (256, log scale).
int VariancePredictorFilterSize { get; set; } Gets or sets the variance predictor filter size (256).
int VariancePredictorKernelSize { get; set; } Gets or sets the variance predictor kernel size (3).
int[] FftKernelSizes { get; set; } Gets or sets the kernel sizes of the FFT blocks’ two convolutions (9 and 1).