Fish Audio

Fish Audio is an AI text-to-speech and voice cloning platform that gives creators and developers expressive synthetic voices at a lower price than many rivals. It grew out of open-source speech research, including the Fish Speech models on GitHub, and now sells a web studio and an API. Its standout feature is fine control over delivery: you write tags like whispering, laughing or a long pause directly into the script.

What is Fish Audio used for?

Fish Audio is used anywhere a script needs a natural-sounding voice, from short videos to real-time agents.

• Voiceovers for YouTube, TikTok and explainer videos
• Narration for stories and audiobooks in the Story Studio editor
• Cloning your own voice from a short sample to read new scripts
• Voices for AI avatars, companions and character apps through the API
• Speech-to-text transcription of recordings
• Separating vocals from background audio and generating sound effects

Who Fish Audio is best for

Fish Audio suits content creators who want emotive voices on a modest budget, and developers building voice agents, avatar products or games who want cheap, fast text-to-speech. Teams that need a tightly curated, licensed voice catalog for regulated brand work may prefer a more enterprise-focused vendor.

Fish Audio pricing

As of October 2026, the Free plan gives 8,000 credits per month, up to about seven minutes of generation, 500 characters per generation and three public voice slots, for personal, non-commercial projects. Plus works out to $11 per month billed annually at $132, with 250,000 monthly credits, private and professional voice slots and Voice Design. Pro costs $75 per month billed annually at $900, with 2,000,000 credits, unlimited voice slots, five professional voice slots and three team seats. Unused credits do not roll over. API billing is separate and pay as you go: the S2.1 Pro model costs $15 per million bytes of text, a free S2.1 Pro tier costs nothing, and transcription costs $0.36 per audio hour.

Is Fish Audio safe and private?

Voice cloning raises consent questions, and Fish Audio's own terms limit commercial use to voices you own. The public library contains many voices uploaded by users, so do not assume a library voice is cleared for commercial work. Clone only your own voice or a voice you have written permission to use, and read the terms of service before publishing.

Fish Audio limitations

• Free plan content cannot be used commercially
• Library voices are uneven in quality and rights status
• Credits reset monthly with no rollover
• The pricing page and FAQ describe annual savings differently, so confirm totals at checkout

Fish Audio vs alternatives

ElevenLabs offers a more polished studio, dubbing and a curated voice marketplace, and is the safer pick for enterprise brand work. Cartesia is built for very low latency in real-time agents. Resemble AI focuses on enterprise cloning with deepfake detection and watermarking. Murf AI and LOVO are aimed at business voiceovers with built-in video editing. Fish Audio wins on price per character and on expressive control through tags.

Tips for getting started with Fish Audio

Write tags into your script sparingly at first, one emotion per sentence, and listen before adding more. For cloning, record a clean sample in a quiet room. If you are a developer, prototype on the free S2.1 Pro API tier and switch to the paid model only when you need its limits.

Fish Audio FAQ

Is Fish Audio free?

Yes, Fish Audio has a free plan with 8,000 credits per month for personal, non-commercial use. Paid plans start at $11 per month on annual billing.

Can I use Fish Audio voices commercially?

Commercial use requires a paid plan, and Fish Audio says paid subscribers can use verified voices that they own for commercial purposes.

Does Fish Audio have an API?

Yes. Fish Audio offers a pay-as-you-go API for text-to-speech, voice cloning and transcription, with Python and TypeScript SDKs.

Fish Audio vs ElevenLabs: which is better?

ElevenLabs has a broader studio and curated voices for enterprise work. Fish Audio is cheaper per character and offers detailed emotion tags.