Open-weight and open-source voice models are advancing rapidly, enabling full-duplex conversations and personalized dialogue on local hardware, marking a shift toward sovereign, low-latency voice interfaces.
NVIDIA released the open-weight Nemotron VoiceChat-11B, an LLM fine-tuned for full-duplex voice conversations. This model signals a shift toward locally executable voice AI, with implications for data sovereignty and hardware efficiency.
An adaptive framework generates dialogues with turn-taking calibrated on a small set of human preferences. It doesn't require massive data – the right annotation aligns interaction to a specific scenario, shifting value from corpus volume to quality ...
Voice AI startup Fish Audio has raised $52 million in a seed round. With over 8 million users and $21 million in ARR within one year, the company leverages an open source offering that enables on-premise deployment, attracting enterprises concerned w...
