Library to reduce latency in voice generations from LLM chat completion streams. - View it on GitHub