phs318u
17 days ago
The GitHub repo URL returns a 404.
Flashlabs releases the world’s first open-source voice cloning model
3 pointsposted 17 days ago
by sangwen2 Comments
kuandriy
17 days ago
The end-to-end speech-to-speech claim is interesting, especially avoiding the ASR→LLM→TTS pipeline, which is where most latency and error compounding happens.