SpeechModel: "best" | "nano"

The speech model to use for the transcription.