Fish Audio is an AI voice platform by Hanabi AI, Inc. offering text-to-speech, voice cloning, and speech-to-text. Voice cloning works from as little as 10 seconds of audio, and the platform hosts a community voice library of over 2,000,000 voices with support for more than 30 languages.
Its current flagship model is Fish Audio S2.1 Pro, an expressive real-time voice model with controllable emotion tags such as angry, sad, excited, and whispering. The platform's open-source lineage comes from the Fish Speech model and the OpenAudio research initiative, with code published on GitHub. Additional tools include Story Studio for audiobook production, a voice changer, audio separation, translation, and sound effect generation.
For voice safety, Fish Audio offers Professional Voice Clone, a verified studio-quality clone built on ownership verification, with optional commercial release and revenue share for the voice owner. Platform policies require users to have rights and consent for any voice they clone. Use cases include video voiceovers, audiobook narration, game and animation character voices, and conversational agents.
Fish Audio is available as a web app and a pay-as-you-go developer API priced at $15 per million UTF-8 bytes with no monthly minimum. A free tier includes 8,000 credits per month with commercial use allowed. Paid plans include Plus ($11/month), Pro ($75/month), and Max ($749/month), with an Enterprise tier offering zero data retention, on-premise deployment, and SOC 2 compliance.