Nuance Labs has raised a $50 million Series A led by Lightspeed Venture Partners. Accel, South Park Commons, NVIDIA and Define Ventures also participated, the Seattle company said in a release.
The funding will support development of Nuance Labs’ full-duplex audiovisual foundation model, expansion of its research team and its first public research preview. The model is designed to accept audio and video, then generate audio and video responses in real time.
Nuance Labs is trying to combine perception and response in one model. It is built to track words, gaze, gestures, tone and timing, then answer with facial and vocal expression. The approach differs from systems that connect separate speech-to-text, language, text-to-speech and animation models.
That architecture addresses a problem in live avatar interactions: conventional multi-step systems can introduce a delay that leaves an avatar appearing frozen while it listens. Nuance Labs is targeting uses including sales, customer service, coaching, professional training, education, language learning and interview practice.
Lightspeed partner Nnamdi Iregbulem said the team has built a single model that follows users in real time and could become a foundational layer for AI products. Co-founder Fangchang Ma said the company’s goal is AI that can understand the ways people express themselves and respond in the moment.
Nuance Labs was founded in 2024 by Fangchang Ma, Edward Zhang and Karren Yang, former Apple PhD researchers who worked on digital personas for Apple Vision Pro. The company is headquartered in Seattle, Washington.
Nuance Labs has not launched a public product. It plans to release its first public research preview later in 2026.
