Fish Audio is a voice AI platform that turns text into speech, clones a voice from a short recording, transcribes audio to text, and exposes the same models through an API. It combines a browser studio with emotion and sound tags, a library of more than 2 million voices, and output in 30+ languages.
The company says 8M+ people use the product, and it publishes an "Audio Turing Test" result of 0.515 across 581 head-to-head comparisons, meaning listeners could not reliably tell its S2.1 Pro model from human speech. Developers can call the hosted API with Python and TypeScript SDKs or self-host open-weight models (fish-speech, S1, S2) under a paid commercial license; HeyGen, Pictoria, Dubbing AI and Plaud are named as API users.
fish.audio · open full screenshot- Starts at
- $11 / month
- Free plan
- Yes
- Integrations
- GPT Image, Seedance, Kling, Veo, Creatify
- API
- Yes
- Category
- Text to Speech
What Fish Audio does
Inline emotion and sound tags
You steer delivery by typing tags into the text itself, such as [angry], [whispering], [excited], [laughing], [sighing] or [long pause]. The developer docs put the direction library at 15,000+ tags that travel with the script rather than as separate parameters.
Voice cloning
Instant voice cloning works from a 30-second audio sample, while professional voice cloning is offered for studio-grade replicas. Cloning quality is one of the criteria customers evaluate alongside latency and language coverage.
Voice library and languages
The voice library lists 2,000,000+ voices you can preview and assign to a script. The platform supports 30+ languages, so one voice can speak more than one language.
Speech to text
Transcription includes multispeaker support, emotion tags and natural-language descriptions of the audio, not just a plain transcript. It sits alongside text to speech in the same product.
Developer API and streaming
A REST call or the Python and TypeScript SDKs return audio, and WebSocket streaming supports sub-second turn-taking for voice agents. The API is billed separately from subscriptions on a pay-as-you-go basis.
Avatar lip-sync
You can generate images and video with GPT Image, Seedance, Kling, Veo and Creatify, then lip-sync the result to a Fish Audio voice for talking avatars, ads and explainers.
How to use Fish Audio
- 1Sign up for the Free plan — no credit card is needed — and open the text-to-speech studio.
- 2Pick a voice from the library or clone one, then paste your script into the editor (up to the character limit of your plan).
- 3Insert emotion and sound tags inline, such as [sad] or [pause], to shape the delivery.
- 4Generate the audio, preview it and download; use Story Studio for longer audiobook projects.
- 5For apps, get an API key, install the Python or TypeScript SDK, and call the text-to-speech endpoint or open a WebSocket stream.
Popular for
Video voiceovers
YouTube creators, advertisers and explainer producers turn scripts into scene-matched narration, swapping tones and adding emotion tags per section.
Audiobook narration
Publishers and indie authors generate chapter-level audio with pacing and emotion control; the site says output is aimed at ACX/Audible specs without a recording booth.
Character and game voices
Game, animation and interactive-story teams clone signature voices or build brand personas and adjust their emotions in the studio or via the API.
Conversational agents
Support bots and virtual agents get low-latency speech with tone tags for helpful or upbeat replies; Dubbing AI uses the text-to-speech API for a real-time voice agent serving its users.
Fish Audio pricing
Freemium- There is a Free plan at $0 / month with 8,000 credits monthly (about 7 minutes of generation), up to 500 characters per generation and 3 public voice slots, for personal, non-commercial use.
- Plus is $15 / month, or $11 / month billed yearly ($132 billed annually), with 250,000 credits, up to 200 minutes, 15,000 characters per generation, unlimited public plus 10 private voice slots, Voice Design, one professional voice slot, enhanced cloning and commercial use.
- Pro is $100 / month, or $75 / month billed yearly ($900 billed annually), with 2,000,000 credits, up to 1,620 minutes, 30,000 characters, unlimited voice slots, 5 professional voice slots, 3 team seats and a 7-day money-back guarantee.
- A minute of generation costs roughly 600–625 credits, unused credits do not roll over, and API usage is billed separately on pay-as-you-go terms.
Pros
- A free tier that needs no credit card, so you can test voices and tags before paying.
- Commercial use is allowed on paid plans, applied to verified voices you own.
- Emotion and sound tags are typed straight into the script instead of set through a separate parameter.
- The API is pay-as-you-go and separate from subscriptions, with an S2.1 Pro free model you can call without credits.
- Open-weight models are available with a paid commercial license, so teams can self-host.
Cons
- The Free plan is limited to personal, non-commercial projects and 500 characters per generation.
- Unused credits reset each billing cycle instead of carrying over.
- Commercial rights on paid plans only cover verified voices you own, not the whole public library.
- Self-hosting the open-weight models is an Enterprise-tier engagement, not a self-serve option.
Fish Audio FAQ
Is Fish Audio free?
Yes. There is a Free plan with 8,000 credits a month, no credit card required, capped at roughly 7 minutes of generation and 500 characters per request. It is for personal, non-commercial use only.
What is Fish Audio used for?
It is used for video voiceovers, audiobook narration, character voices, conversational chatbots, avatar lip-sync and speech-to-text transcription. The same models are available through the web studio and the API.
How much audio do I need to clone a voice?
Instant voice cloning works from 30 seconds of audio. Professional voice cloning is a separate option for studio-grade replicas.
Does Fish Audio have an API?
Yes. It offers a REST endpoint, WebSocket streaming and official Python and TypeScript SDKs. API usage is billed separately from subscriptions, pay-as-you-go, and the S2.1 Pro free model can be called without credits.
Can I use generated audio commercially?
Premium subscribers can use verified voices they own for commercial purposes. Free plan output is limited to personal, non-commercial projects.
How many credits does a minute of audio cost?
Roughly 600 to 625 credits per minute of generation, according to the pricing page.
Written from fish.audiofish.audio/plan/fish.audio/customers/fish.audio/developers/· reviewed by a person · updated Nov 5, 2026