gpt-4o-tts1 allows even more individual sounding voices with detailed instructions, like 'sound: whisper very slow'.