SpeechModel: "tts-1" | "tts-1-hd" | "gpt-4o-mini-tts"