[mythhabit]

If I'm understanding the specs correctly, it is basically a LLM, but for audio. So it requires some serious power to encode it, because it is using the latest AI hype to achieve the result.