Voice-to-voice chatbot update
Signal
72
Hype
25
In three linesReal-time local voice chatbot using Qwen3.5-397B (Unsloth UD-Q3_K_XL), Whisper-small STT, and Orpheus Q4_K_XL TTS with custom SNAC decoder on ONNX. Interruptible with context preservation, 21.3 GB VRAM max on 24GB GPU, bf16 KV cache at 131k tokens. GitHub code coming soon.Read source
Your take?
Summary generated by Claude — human-verified