Refactor voiceline system to use audioclips instead of procedurally generating audio at runtime, todo now make cooldowns use voiceline type not exact text as id