v1-mini

LiveKit released two end-of-turn detection models that process speech directly instead of relying on transcripts, fusing semantic and acoustic understanding. The v1 model achieves top results across 1…