Fish Audio Review: Features, Pricing and Alternatives
Fish Audio review as of October 2026: S2.1 Pro and Drama 3 models, plan prices, pay-as-you-go API rates, voice cloning, licensing, and where it falls short.
Fish Audio is a text to speech and voice cloning platform with S2.1 Pro and Drama 3 models. Its free plan includes 8,000 credits a month, paid plans start at $15 a month ($11 billed yearly), and the S2.1 Pro API costs $15 per 1M UTF-8 bytes. Check the terms before using free output commercially.
This review covers what Fish Audio offers as of October 2026, what each plan and API rate costs, how its licensing works, and where it has limits. All prices come from Fish Audio's own pages, checked on 2026-10-17.
What is Fish Audio?
Fish Audio is a voice AI company with a text to speech product, a developer API and open-weights models. It says it has more than 8 million users and more than 2 million voices in its community library. It also announced US$52 million in seed funding on 2026-07-27, led by Coreline Ventures and Capital Today. Those numbers come from the company, and we have not verified them.
The company reports that enterprise and developer customers make up about two-thirds of its revenue. Treat that as a vendor statement too. For a buyer, the practical questions are the plan you need, the license on the output, and whether the model fits your languages.
Fish Audio features
The Fish Audio navigation lists a wide set of tools. The main ones are:
- Text to speech, with expressive tags such as [angry], [whispering], [laughing] and [pause] in the script
- Speech to text, with transcription models priced per audio hour
- Voice cloning and Voice Design, where Voice Design comes with the Plus plan
- Voice changer, audio separation and audio translation
- Story studio, sound effects and avatar lipsync
- Image and video generation, and an MCP server
The web text to speech box accepts up to 30,000 characters of text. On the Enterprise tier, Fish Audio lists zero data retention, on-premise deployment, SOC2 compliance and HIPAA-compliant configurations. Those are vendor statements from its plan and blog pages.
Fish Audio models: S2.1 Pro and Drama 3
Fish Audio's model history is useful for choosing a version. The company open-sourced S2 on 2026-03-09. It says S2 was trained on more than 10 million hours of audio across about 50 languages, and the release includes weights, fine-tuning code and an SGLang inference engine. Fish Audio reports word error rates of 0.54% in Chinese and 0.99% in English on the Seed-TTS Eval benchmark. Those are vendor-reported results from the S2 open-source announcement.
S2.1 Pro is the model that the API lists at $15 per 1M UTF-8 bytes. Drama 3 is the newer release. It launched on 2026-10-13, is available in the web app, the iOS and Android apps and the API, and supports 83 languages according to Fish Audio. Fish Audio also says Drama 3 scores 93.2% on pronunciation robustness, against 88.1% for S2.1 Pro, using the Artificial Analysis test. That is Fish Audio's own report of the test, and it is not our measurement. Read the Drama 3 launch post for the full category breakdown.
Fish Audio pricing and plans
The plan prices below come from the Fish Audio plan page, checked on 2026-10-17.
| Plan | Monthly | Billed yearly (per month) | Credits per month | Stated maximum length | Professional voice slots | |---|---|---|---|---|---| | Free | $0 | Not applicable | 8,000 | Up to 7 minutes | None listed | | Plus | $15 | $11 ($132 a year) | 250,000 | Up to 200 minutes | 1 | | Pro | $100 | $75 ($900 a year) | 2,000,000 | Up to 1,620 minutes | 5 | | Max | $999 | $749 ($8,988 a year) | 25,000,000 | Up to 6,250 minutes | 15 | | Enterprise | Custom | Custom | Custom | Custom | Custom |
A few details sit beside the table. The Free plan allows up to 500 characters per generation and 3 public voice slots. Plus adds unlimited public voices, 10 private voices and access to Voice Design. Pro includes 3 team seats and a 7-day money-back guarantee. Max includes 10 seats. The plan page describes annual billing as three months free.
Fish Audio's FAQ says that about 600 to 625 credits equal one minute of generation. Credits reset each month, and unused minutes do not roll over. The plan cards use their own minute figures, so use those when you compare plans.
Fish Audio API pricing and rate limits
The API is pay-as-you-go. Fish Audio says there are no subscription fees or monthly minimums for API access. The rates below come from the Fish Audio pricing and rate limits page.
| Model or service | Price | Unit | |---|---|---| | s2.1-pro and s2-pro | $15.00 | Per 1M UTF-8 bytes | | s2.1-pro-free | $0.00 | Per 1M UTF-8 bytes, for an initial period | | transcribe-1-pro and transcribe-1 | $0.36 | Per audio hour | | voice-design-1 | $0.01 | Per successful request | | Drama 3 (list price) | $50 | Per 1M characters |
Three notes apply to this table:
- The s1 model is deprecated and retires on 2026-12-31. Requests then go to s2.1-pro at the same price.
- Drama 3 is not in the docs pricing table. Its launch post gives a list price of $50 per 1M characters, with a promotion of $10 per 1M characters until 2026-11-12. Do not apply the $15 per 1M bytes rate to Drama 3.
- The s2.1-pro-free model is listed at $0. Fish Audio describes free API access as an initial period, with advance notice of changes. Do not plan a product around it being permanent.
Concurrency depends on how much you have prepaid. Starter, under $100, allows 5 concurrent requests. Elevated, at $100 or more, allows 15. High Volume, at $1,000 or more, allows 50. Enterprise is custom.
The docs note that 1M UTF-8 bytes is about 180,000 English words, or about 12 hours of speech. That gives a derived figure of about $0.02 per minute of English speech. The figure applies only to English, and only as arithmetic on the list price.
Can you use Fish Audio commercially?
The answer depends on the plan and the kind of voice you use. Fish Audio's plan FAQ says paid subscribers can use verified voices they own for commercial purposes. It says free plan users can use generated content only for personal, non-commercial projects. The free plan card lists commercial use among its features, which conflicts with the FAQ. Because of that conflict, check the terms for your specific use.
The open-weights models have their own license. Drama 3 is released under the Fish Audio Community License v1.0. It is free for research and non-commercial use. Commercial use is free if you and your affiliates have under US$500,000 in annual revenue. You need a license above that threshold, or if you offer the model as a TTS, voice-cloning or inference service or API.
Fish Audio pros and cons
Strengths, based on the pages we checked:
- Published API pricing with pay-as-you-go billing and no monthly minimum
- Open-weights releases for S2 and Drama 3
- A broad set of tools, including speech to text, voice changing and sound effects
- Plan pages that state maximum generation lengths, so you can estimate capacity
Limits to plan around:
- Free API access is described as an initial period, and the free plan's commercial status is unclear from the page.
- Promotional prices, such as the Drama 3 API discount and creator plan offers, are time-limited.
- Fish Audio's pages give conflicting sample lengths for voice cloning, so check current guidance before you record.
- Language counts differ by product. Drama 3 is listed at 83 languages, and the homepage cloning line says 30+.
Fish Audio alternatives
Fish Audio is one option among several. Our ElevenLabs alternatives guide compares it with Cartesia, Murf and others. For voiceover-focused teams, Murf AI alternatives covers how tools meter usage.
OndaLabs is an AI voice studio. It offers text to speech with 30 prebuilt voices that work in every supported language, two-speaker dialogue, speech to text, custom voices and image and music generation. Custom voices can be cloned from a recording you have permission to use, or designed from a text description. The free plan includes 2,000 credits a month, and commercial use starts on the Starter plan at $9 a month. You can compare plans on the OndaLabs pricing page or design a voice in the custom voices area.
Fish Audio publishes an API, open-weights models and a much larger voice library than OndaLabs. OndaLabs does not make those claims. If you need an API or self-hosted weights, Fish Audio covers more ground. If you want a studio for narration, dialogue and transcription under one credit balance, compare the two side by side with the plan pages.
Frequently asked questions
What is Fish Audio?
Fish Audio is a voice AI company offering text to speech, voice cloning, speech to text and related tools. Its models include S2.1 Pro, served through the API, and Drama 3, which launched on 2026-10-13. Fish Audio also publishes open-weights models under its own license.
How much does Fish Audio cost?
On Fish Audio's plan page, as of October 2026, the Free plan is $0, Plus is $15 a month or $11 billed yearly, Pro is $100 a month or $75 billed yearly, and Max is $999 a month or $749 billed yearly. The API is billed separately on a pay-as-you-go basis.
Is Fish Audio free to use?
Fish Audio has a free plan with 8,000 credits a month, and the free API model s2.1-pro-free is listed at $0 for an initial period. Free plan output is for personal, non-commercial projects according to its FAQ. The free API status could change, so check the current terms.
Can I use Fish Audio voices commercially?
Paid subscribers can use verified voices they own for commercial purposes, according to Fish Audio's plan FAQ. Free plan users are limited to personal, non-commercial projects. Open-weights models have a separate license, which requires a fee above US$500,000 in annual revenue.
Is Fish Audio open source?
Fish Audio has released open weights for S2 and Drama 3. S2 was released on 2026-03-09 with weights, fine-tuning code and an inference engine. Drama 3 is released under the Fish Audio Community License v1.0, which limits commercial use above a US$500,000 annual revenue threshold.