Full-duplex spoken dialogue research: fine-tuning a speech model (LFM3-Audio-2B, QLoRA) toward an offline voice assistant that listens while it talks, on a 4 GB laptop GPU. Paused; code, frozen test definitions, results and a review paper.
speech voice-assistant conversational-ai full-duplex spoken-dialogue-system voice-ai turn-taking speech-model speech-to-speech llama-cpp qlora offline-ai cuda-graphs spoken-dialogue lfm3-audio
-
Updated
Nov 11, 2026 - Python