SpeechModel: "best" | "nano" | "slam-1"

The speech model to use for the transcription.