Property SuppressNonSpeechTokens
SuppressNonSpeechTokens
Gets or sets whether to suppress non-speech tokens during transcription.
public bool SuppressNonSpeechTokens { get; set; }
Property Value
Examples
var engine = new SpeechToText(model);
engine.SuppressNonSpeechTokens = false; // Disable non-speech token suppression
var result = engine.Transcribe(new WaveFile("audio.wav"));
Remarks
When enabled, the model suppresses tokens that do not represent actual speech, such as filler sounds, background noise artifacts, and repetitive patterns. This helps reduce erroneous repetitions and improves overall transcription quality.
Defaults to true.