HuBERTOptions

Options & Configuration · AiDotNet.Audio.Foundations

Configuration options for the HuBERT (Hidden-Unit BERT) self-supervised speech model.

Inheritance: ModelOptions → HuBERTOptions

For Beginners

HuBERT learns to understand speech by listening to millions of hours of audio. It predicts hidden “units” in speech (like phonemes) without any labels. After pre-training, it can be fine-tuned for tasks like transcription or speaker ID.

How It Works

HuBERT (Hsu et al., 2021, Meta) learns speech representations by predicting masked discrete speech units derived from clustering. It achieves strong performance on speech recognition, speaker verification, and emotion detection when fine-tuned.

Constructors

Constructor Summary
HuBERTOptions()

Properties

Property Summary
OnnxModelOptions OnnxOptions { get; set; } Gets or sets the ONNX runtime options.
bool StableLayerNorm { get; set; } Whether to build HuBERT-Large’s layout: a layer-norm feature encoder with conv bias and a stable (pre-norm) Transformer.
double ActivationDropout { get; set; } Dropout inside each feed-forward block (facebook/hubert-base-ls960: 0.1).
double DropoutRate { get; set; } Gets or sets the dropout rate.
double LayerDrop { get; set; } Probability of skipping each Transformer layer during training (facebook/hubert-base-ls960: 0.1).
double LearningRate { get; set; } Gets or sets the peak learning rate of the pre-training schedule (5e-4, HuBERT Base’s; Hsu et al.
double MaskTimeProbability { get; set; } SpecAugment time-mask probability during training (facebook/hubert-base-ls960: 0.05).
int FeatureEncoderDim { get; set; } Gets or sets the CNN feature encoder output dimension.
int FeedForwardDim { get; set; } Gets or sets the feed-forward dimension.
int HiddenDim { get; set; } Gets or sets the transformer hidden dimension.
int NumAttentionHeads { get; set; } Gets or sets the number of attention heads.
int NumLayers { get; set; } Gets or sets the number of transformer layers.
int SampleRate { get; set; } Gets or sets the sample rate in Hz of the audio passed to the model.
string ModelPath { get; set; } Gets or sets the path to the ONNX model file.
string Variant { get; set; } Gets or sets the model variant (“base” or “large”).