Fish Audio, a Palo Alto startup focused on AI-generated voice models, has raised $52 million in a seed funding round led by Coreline Ventures and Capital Today, with participation from multiple other investors. Since its launch last year, Fish Audio has attracted over 8 million users through open source and hosted versions of its voice models. The company currently generates $21 million in annual recurring revenue, reflecting strong market demand for its technology.

The company was founded by former Nvidia researcher Shijia Liao, who initially created the voice-generation model out of frustration with the lack of expressiveness in synthetic voices. His project, open-sourced as Fish Speech on GitHub, now boasts more than 31,000 stars and is widely adopted by indie developers, game designers, and creative professionals. Fish Audio has released a total of five models, including four for speech generation and one for speech-to-text conversion.

Fish Audio’s product offerings include both open-source models and paid subscription plans that unlock additional features such as voice cloning and higher usage limits. The latest model, S2.1 Pro, is exclusively accessible through the company’s paid API. The company also provides an enterprise solution tailored to organizations, with existing customers like HeyGen, Sanas, and LiveKit leveraging its technology for AI avatars, gaming characters, and voice agents.

According to CEO Rissa Cao, Fish Audio aims to address a wide range of use cases by delivering voice models that meet the varying demands of creators and enterprises. This includes providing realistic voices for AI avatars, expressive speech for gaming characters, and natural, low-latency voices suitable for customer support and sales operations. With significant new funding, Fish Audio plans to expand its capabilities and improve the quality and adaptability of its AI voice technology.