Fish Audio Community License Agreement 1.0

Free for research and non-commercial use, and free for commercial use while you and your affiliates earn less than US$500,000 a year. Above that, or to offer the model to others as a text-to-speech, voice-cloning or inference service at any size, you need a separate written license from Fish Audio. You must show "Built with Fish Audio", you may not clone a real person's voice without consent, and you may not use the model or its audio to train other speech or foundation models.

  • Conditional (Conditional)
  • Not an Open Source Initiative-approved license
  • Verified

Commercial use is allowed, but with extra conditions: a revenue cap, attribution, or a separate deal for large hosting businesses.

What you can do

  • Use, copy, modify, fine-tune, quantize and merge the model for research, personal or evaluation purposes at no cost
  • Use it commercially while the total yearly revenue of you and your affiliates is under US$500,000 (the "Community Commercial License")
  • Generate speech for your own content or product (narration, dubbing, game or app characters, an assistant's replies), and deliver audio to clients as an agency, studio or freelancer as long as they get no self-serve access
  • Run a free, non-commercial research demo, such as a Hugging Face Space, that is not offered as a production service
  • Own your derivative models and the audio you generate, as far as the law allows

What to watch out for

  • Commercial use above US$500,000 in yearly revenue (you and your affiliates together) needs a separate written license; you must tell Fish Audio within 30 days of crossing the threshold, and the community terms keep applying for 90 more days
  • A "Hosted Service", meaning an API, SDK, hosting service or website whose main purpose is text-to-speech, voice generation or voice cloning for third parties, is not licensed at any revenue level
  • If you deploy the model for a client as a contractor, integrator or on-premises vendor, that client's own revenue must also be under US$500,000
  • If you distribute or make available the model, or a product or service that uses it, you must show "Built with Fish Audio" on a related website, interface or documentation, and copies you hand to others need the agreement and a NOTICE file crediting 39 AI, Inc.
  • You may not imitate an identifiable person's voice without consent, impersonate people or organizations to deceive, defraud or harass, or remove watermarks or provenance signals
  • You may not use the model, its outputs or its derivatives to train or distill any foundation model or any speech, audio or voice generation model; only detectors of synthetic audio are exempt. Breaking any of these use restrictions ends the license automatically, with no notice or cure period
  • The Hugging Face metadata calls the license "fish-audio-research-license", but the LICENSE file is the Fish Audio Community License Agreement, and the older S2 weights stay under the Fish Audio Research License
  • The agreement is governed by US and California law, and your license ends if you sue Fish Audio over intellectual property claims about the materials or their outputs

Conditions that apply

  • Attribution or model name must be shown
  • Free commercial use only below a revenue cap
  • Separate deal needed to sell hosted access above a threshold
  • Acceptable-use policy applies

This is a plain-language summary, not legal advice. The license text is the only authoritative source.

Read the full license text

Models released under this license

Model Released Parameters Context License Smallest variant needs
Drama 3 Fish Audio 35.9B 3B active – Fish Audio Community 1.0 (Conditional) 25 GB Gaming GPU