Fish Audio

Fish Audio is a voice AI platform for expressive text-to-speech and fast voice cloning, able to clone a voice from as little as a 15-second sample and support zero-shot cloning across many languages via its S2 model. It offers dozens of emotion and tone tags and low-latency streaming for conversational AI. Uniquely, its core OpenAudio and Fish Speech models are open source, allowing self-hosted, no-per-generation-cost deployment alongside a hosted API and web app. Founded in 2025 and headquartered in Palo Alto, it raised a $52 million seed round in mid-2026.

Fish Audio is a fast-growing voice AI company that produces expressive text-to-speech and voice cloning. Its models can clone a voice from a sample as short as 15 seconds, capturing timbre, pacing, and style, and its S2 model supports zero-shot cloning across a large set of languages. A hallmark is granular expressiveness, with dozens of emotion and tone tags plus low-latency streaming aimed at conversational AI. What sets Fish Audio apart is that its core models, including Fish Speech and the OpenAudio S1 and S2 releases, are published as open-source repositories. That allows developers to self-host and run generation locally without per-use cost, while a hosted platform and API provide a managed option with a public voice library. The company reports strong benchmark rankings for naturalness and expressiveness. Founded in 2025 by former NVIDIA researcher Shijia Liao and headquartered in Palo Alto, Fish Audio raised a $52 million seed round in mid-2026 led by Coreline Ventures and Capital Today, and reports rapid growth to millions of users. It is a compelling choice for creators, developers, and enterprises wanting flexible, affordable, expressive voice generation.