Nuance Raises $50M for Unified Audiovisual AI Model

Key Takeaways

1Nuance, founded in Seattle by three former Apple researchers, raises $50 million in a Series A round led by Lightspeed

2The startup is developing a unified audio-video AI model and is collecting its own data, without relying on web data

3A public preview of the model is scheduled for this year, targeting use cases in coaching, learning, and business functions

4A first round of $10 million was led by Accel last July

Why it matters : Nuance aims to surpass the limitations of current chatbots by focusing on natural and responsive conversations, with proprietary architecture and data.

The Seattle-based startup, founded last year by three former Apple researchers, aims for a conversational AI that enables face-to-face interactions with no perceptible latency. It has raised $50 million in a Series A round led by Lightspeed, is preparing a public preview of its model later this year, and is expanding its team to eight people after an initial $10 million round led by Accel last July.

A Preview Announced This Year and a Growing Team

The company is still in the pre-product phase and plans to open a public research preview of its model later this year. Fangchang Ma indicates that the cash flow is being used to develop a team of eight, while the company invests in recruitment for R&D. The next announced step is to hire go-to-market specialists to refine pricing and monetization strategies. Founded last year, Nuance had previously raised $10 million last July in a round led by Accel.

Betting on a Single Audio-Video Model and Proprietary Data

Fangchang Ma describes a unique architecture, with a single system and model taking audio and video as both input and output. He criticizes current setups that combine speech recognition, language models, and synthesis, which he sees as sources of latency and loss of nuances such as reactions and micro-expressions, noting that some avatars do not respond to listening or behave randomly. According to him, the required data cannot be found online: Nuance collects it internally or through partners, aiming for bilateral recordings with visible participants, separate audio tracks, and natural exchanges. The three co-founders, who have backgrounds in research at Apple, claim a long-term approach despite its complexity.

Le brief IA que lisent les pros

Le brief IA que les pros lisent chaque soir

Les 7 actus IA du jour, décryptées en 5 min. Gratuit.

Inclus dès l'inscription : notre sélection des meilleurs guides & comparatifs IA.

Choisissez votre rythme

Gratuit · Pas de spam · Désabonnement en 1 clic

$50 Million Series A, Led by Lightspeed

Nuance has closed a $50 million Series A round led by Lightspeed Venture Partners. The round also includes Accel, Nvidia's NVentures, and South Park Commons. This funding, aimed at developing an audiovisual AI model, comes as the company is based in Seattle.

Use Cases Considered from Coaching to Business Functions

The team led by CEO and co-founder Fangchang Ma aims to create chatbots capable of real-time responses and replicating human subtleties for face-to-face exchanges without delay. Ma envisions a connection to a video call with an AI for practicing interviews, making presentations, or learning a language. On the business side, Nuance plans to offer services for automated interviewers, sales representatives, and customer support.