SqueezeformerOptions

Options & Configuration · AiDotNet.SpeechRecognition.ConformerFamily

Configuration options for the Squeezeformer speech recognition model.

Inheritance: ModelOptions → SqueezeformerOptions

How It Works

Squeezeformer (Kim et al., 2022) improves Conformer with a temporal U-Net structure, micro-macro design (pre-norm instead of post-norm), and depthwise separable downsampling for efficient computation with better accuracy.

Constructors

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

Properties

Property Summary
OnnxModelOptions OnnxOptions { get; set; } Gets or sets the ONNX Runtime settings used when the model is loaded from an ONNX file.
bool UseLayerNormalization { get; set; } Gets or sets whether the native encoder normalises with LayerNorm rather than BatchNorm.
double DropoutRate { get; set; } Gets or sets the dropout probability of the native layers during training.
double LearningRateDecayPower { get; set; } Gets or sets the polynomial decay power after the peak-hold phase.
double PeakLearningRate { get; set; } Gets or sets the PEAK learning rate for the Noam-annealed schedule.
double WeightDecay { get; set; } Gets or sets AdamW’s decoupled weight decay.
int EncoderDim { get; set; } Gets or sets the hidden width of the encoder.
int FeedForwardExpansionFactor { get; set; } Gets or sets the inner width of each feed-forward block as a multiple of the encoder width.
int MaxAudioLengthSeconds { get; set; } Gets or sets the longest clip, in seconds, of the reference configuration.
int NumAttentionHeads { get; set; } Gets or sets the number of heads in each attention layer.
int NumEncoderLayers { get; set; } Gets or sets the number of encoder layers.
int NumMels { get; set; } Gets or sets the number of mel filterbank channels in the input features.
int PeakHoldSteps { get; set; } Gets or sets how many optimizer steps hold the peak learning rate after warmup.
int SampleRate { get; set; } Gets or sets the sample rate, in Hz, that input audio is expected at.
int VocabSize { get; set; } Gets or sets the number of output tokens the model predicts over.
int WarmupSteps { get; set; } Gets or sets the warmup steps for the extended Noam learning-rate schedule.
string Language { get; set; } Gets or sets the language code (for example “en”) reported for a transcript when none is passed or detected.
string ModelPath { get; set; } Gets or sets the path of the pretrained ONNX model.
string[] Vocabulary { get; set; } Gets or sets the token strings, indexed by token id, that native transcripts are spelled with.