OpenAudio / Fish Audio

Fish Audio builds text-to-speech, voice cloning and speech-to-text tools for creators, developers and enterprises. Users can turn scripts into video narration and audiobooks, create character voices for games, and add spoken responses to customer-support agents. Its web tools and APIs provide control over emotion and delivery, while enterprise deployment options include self-hosting for teams that want control over their infrastructure and data.

Founded in 2025 by CEO Rissa Cao and Chief Scientist Shijia Liao, the company grew from Liao’s Fish Speech project, initially trained on a single gaming GPU. Its research connects speech generation with control over how words are spoken. Fish Audio S2 supports multiple speakers and conversational turns, with delivery guided by natural-language instructions. The team released model weights, fine-tuning code and an SGLang-based streaming inference engine, giving developers tools to adapt and serve the system.

In July 2026, Fish Audio announced $52 million in seed funding, led by Coreline Ventures and Capital Today. The company reported reaching $21 million in annual recurring revenue from zero in its first year, with more than eight million users and 22 staff. Developers and enterprises together accounted for two-thirds of revenue, alongside a community library containing more than two million voice models.