Nuance, a Seattle-based startup founded last year, has raised a $50 million Series A led by Lightspeed Venture Partners to develop AI models that can conduct more natural, emotionally aware face-to-face conversations. Cofounder and CEO Fangchang Ma said current voicebots and AI avatars typically connect voice-to-text, a large language model, and text-to-voice systems, creating latency and missing reactions or microexpressions. Nuance is pursuing a single model with audiovisual input and output instead. The pre-product company gathers training data in-house or through data partners because it says suitable audiovisual conversations cannot be scraped from the internet. It plans to open a public research preview later this year and eventually target consumers and businesses with applications including interview practice, presentations, language learning, AI interviewers, sales representatives, and customer support. The round follows a $10 million seed investment led by Accel in July last year. Nuance has used the funding to expand its eight-person team, is hiring for research and development, and plans to add go-to-market specialists to establish pricing and monetization.