disfluencies parameter controls whether this speech is kept in the transcript.
Defaults by model
The filler word default depends on the speech model. AssemblyAI-STT-3.7 keeps filler words by default. Universal-3.5 Pro, Universal-3 Pro, and Universal-2 remove them by default. To get the same behavior on every model, setdisfluencies explicitly.
If
speech_models includes assemblyai-stt-3-7 in any position, disfluencies defaults to true for the whole request, even when another model is listed first or transcribes the audio. For example, ["universal-2", "assemblyai-stt-3-7"] defaults to true. If speech_models doesn’t include assemblyai-stt-3-7, the default comes from the first model in the list.
Include verbatim speech
Setdisfluencies to true in your transcription config to keep verbatim speech. You need to set it when speech_models doesn’t include assemblyai-stt-3-7, for example ["universal-2"].
What disfluencies: true preserves depends on the model:
Remove filler words on AssemblyAI-STT-3.7
AssemblyAI-STT-3.7 keeps filler words unless you turn them off. Setdisfluencies to false to remove them:
Supported languages
Verbatim output (disfluencies: true) is supported in the following languages: