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

Fish Audio, officially registered as Hanabi AI Inc., announced today it has closed a $52 million seed funding round to establish voice as the default interface for every AI model. The round was led by Coreline Ventures and Capital Today, with additional participation from 359 Capital, Play Time, HF0, 645 Ventures, Parable, Carya Venture Partners, Alphalist Partners, and several unnamed angels.

The company's founding story is rooted in frustration with early AI voice quality. Co-founder and Chief Scientist Shijia Liao, who previously worked as a video researcher at Nvidia and is a lifelong fan of Japanese anime, grew tired of the flat, monotonously robotic synthetic voices produced by early AI models. He decided to train his own voice AI models using nothing more than a single GPU housed in his bedroom laptop. Despite these severe hardware constraints, Liao developed the initial text-to-speech and voice cloning models that evolved into Fish Speech, an open-source project that rapidly gained traction on GitHub, accumulating over 31,000 stars. The project caught the attention of indie developers, content creators, and video game designers hungry for more expressive voice generation capabilities.

Since its official launch in 2023, Fish Audio has evolved into one of the most comprehensive voice AI platforms available. The platform is built around eliminating the lifeless synthetic audio of earlier models by providing granular, word-level emotion controls powered by over 15,000 natural language prompts. This allows developers to fine-tune the exact tone, inflection, and pacing of AI-generated voices. The platform can clone a voice from a mere five-second audio clip in less than 15 seconds, natively supports 83 languages, and its flagship S2.1 Pro model outperformed competitors in blind listening tests, with 67% of listeners preferring Fish Audio's voice outputs. These capabilities have driven the company to over eight million users, with annual recurring revenue now exceeding $21 million.

Initially targeting video game developers and content creators, Fish Audio has expanded into regulated industries such as healthcare and financial services. The company now offers secure on-premises deployments with zero-data retention and HIPAA compliance to protect customer privacy. This enterprise expansion positions Fish Audio as a formidable challenger to better-known voice AI startups like ElevenLabs, which recently raised $500 million in a round valuing the company at $11 billion.

Chief Executive Rissa Cao, who co-founded the company alongside Liao, said the mission was to make human-sounding AI voices accessible to everyone. "We make high-quality, human-sounding voices available to every user, from beginner creatives to million-dollar enterprises, so communication is not only more efficient, but more trustworthy," he said. "We've always believed that if we kept making the models better, people would notice. Eight million of them did."

With the fresh capital, Fish Audio plans to expand its model lineup significantly. Beyond its current text-to-speech capabilities, the company intends to build a full audio-native stack composed of voice-native large language models and real-time speech-to-speech translation tools. It will also invest in an enterprise sales team and expand developer tooling with additional API integrations through partners such as Retell AI and LiveKit. To accelerate platform adoption, Fish Audio will make its flagship S2.1 Pro model available free of charge to every developer through its official API, beginning at the end of August.

Coreline Ventures' managing partner Osuke Honda explained his decision to invest, saying: "In its short history, Fish Audio has built an unbeatable track record of pushing the envelope on performance, multilingual support, emotional expression and cost. All factors that have quickly made Fish Audio the default choice for creators, developers, and now enterprises globally, and we expect them to continue to lead the way."