Fish Audio
AI voice platform from the Fish Speech open-source project — 83+ languages, voice cloning, $52M seed.
About
Fish Audio is an AI voice platform built for speed and control, founded in 2025 in Palo Alto. It grew out of Fish Speech, one of the most-starred open-source voice projects on GitHub (31,000+ stars), into a platform serving more than 8 million creators, developers, and enterprises including HeyGen, Retell, LiveKit, and Sanas. Its open models deliver expressive real-time text-to-speech, voice cloning, and voice agents in more than 83 languages, and can be self-hosted for teams that want full data control.
The honest picture: the viral "clone any celebrity voice for free" angle raises obvious ethical and legal questions — users must respect voice rights. The company raised $52M in seed funding in mid-2026 and is expanding fast, which means pricing and packaging are still in flux. Enterprise features like HIPAA compliance and on-premises deployment exist but are gated behind sales conversations.
What it can do
- Expressive text-to-speech in 83+ languages
- Voice cloning from ~15 seconds of audio
- Voice agents and speech-to-speech
- Self-hostable open models
- 2M+ community voice library
Setup
Sign up at fish.audio for the web playground, or use the API — the S2.1 Pro model has been offered free for developers via API in promotional windows.
Prerequisites
A Fish Audio account; API key for programmatic access.
Use cases
Works with
Trust score — 8/10
Scored 8/10 (verified tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 53; Trending score 85/100 from researcher assessment.
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.