Fish Audio raises $52M to build human-sounding AI voices

Fish Audio emerged from a classic startup origin story: a single engineer frustrated with poor technology who decided to build a better solution. Co-founder Shijia Liao, a video researcher at Nvidia who is a lifelong anime fan, began training voice AI models on a single GPU in his bedroom laptop. His open-source project Fish Speech quickly gained traction on GitHub, amassing over 31,000 stars as indie developers, content creators, and game designers sought more expressive voice generation tools. Since launching in 2023, the company has grown to serve over eight million users with annual recurring revenue exceeding $21 million—remarkable for a company that started as a weekend project.

The $52 million seed round reflects strong investor confidence in voice AI as a fundamental interface for AI systems. Coreline Ventures' managing partner Osuke Honda explicitly framed the investment around voice becoming "the default interface for AI systems," suggesting investors see voice as an increasingly critical competitive layer. Fish Audio's performance metrics support this view: its S2.1 Pro model outperformed competitors in blind listening tests, with 67% of listeners preferring its output—a meaningful differentiator in a market where tone and naturalness directly impact user experience and adoption.

Fish Audio now competes directly with better-known startups like ElevenLabs, which recently raised $500 million at an $11 billion valuation. The company's pivot to enterprise markets—offering secure on-premises deployments with HIPAA compliance for healthcare and financial services—suggests a strategy to capture higher-margin, regulated customers where data privacy is non-negotiable. Free API access to S2.1 Pro starting at end of August may be a tactic to accelerate developer adoption and lock in users before scaling the sales team and expanding into adjacent products like voice-native large language models and speech-to-speech translation.