WhisperOptions

Options & Configuration · AiDotNet.Audio.Whisper

Configuration options for the Whisper speech recognition model.

Inheritance: ModelOptions → NeuralNetworkOptions → ModelHyperparameterOptions → AudioNeuralNetworkOptions → WhisperOptions

For Beginners

Whisper comes in different sizes (tiny to large). Smaller models are faster but less accurate. Larger models are more accurate but slower.

  • Tiny: ~39M parameters, fastest, good for quick transcription
  • Base: ~74M parameters, balanced speed/accuracy
  • Small: ~244M parameters, good accuracy
  • Medium: ~769M parameters, high accuracy
  • Large: ~1.5B parameters, best accuracy, slow

How It Works

Whisper is a speech recognition model developed by OpenAI that can transcribe audio in multiple languages and perform translation.

Constructors

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

Properties

Property Summary
OnnxModelOptions OnnxOptions { get; set; } Gets or sets the ONNX execution options.
WhisperModelSize ModelSize { get; set; } Gets or sets the model size to use.
bool Translate { get; set; } Gets or sets whether to translate to English.
double Temperature { get; set; } Gets or sets the temperature for sampling.
int BeamSize { get; set; } Gets or sets the beam size for beam search decoding.
int MaxAudioLengthSeconds { get; set; } Gets or sets the maximum length of audio to process in seconds.
int MaxTokens { get; set; } Gets or sets the maximum number of tokens to generate.
int NumMels { get; set; } Gets or sets the number of mel filterbank channels.
int SampleRate { get; set; } Gets or sets the sample rate expected by the model.
string Language { get; set; } Gets or sets the language code for transcription (e.g., “en”, “es”, “fr”).

Methods

Method Summary
void Validate() Throws if a value this model requires has been left unset or is not positive.