Ex-Apple Researchers’ Nuance Raises $50M to Build More Emotionally Intelligent AI Avatars

Seattle-based Nuance has raised a $50 million Series A to build AI models aimed at making avatars and voicebots better conversationalists. The round was led by Lightspeed Venture Partners, with participation from Accel, Nvidia’s NVentures, South Park Commons, and Define Ventures.

The company was founded last year by Fangchang Ma and two cofounders, all former Apple researchers. The new funding follows a $10 million seed round led by Accel in July last year.

Nuance is targeting a familiar pain point in conversational AI: latency, stiff reactions, and missing human cues. Ma said many existing voicebots and AI avatars combine voice-to-text, a large language model, and text-to-voice generation, which can create delays and lose subtle signals like reactions and microexpressions.

The company’s proposed answer is a single system with audio and video in, and audio and video out. That approach is described as more technically challenging, but central to Nuance’s plan for more natural face-to-face AI interactions.

Ma said the training data Nuance needs cannot be scraped from the internet. Instead, the startup collects data in-house or through data partners.

The target data is audiovisual material from both sides of a natural conversation, with both participants visible and cleanly separated soundtracks. That focus signals how much Nuance’s model depends on context, timing, and human expression rather than text alone.

Nuance is still pre-product, but Ma described future experiences such as practicing interviews, giving presentations, or learning a new language through a video call with AI. The company also plans to sell to businesses, with possible use cases including AI interviewers, sales reps, and customer support.

Nuance plans to open a public research preview later this year. The startup is using its funding to expand its eight-person team, invest in research and development, and hire go-to-market experts to refine pricing and monetization.