GOD.EXEA god's-eye view of the AI universe
2026Drag to orbit · right-drag to pan · scroll to zoom
GOD.EXE is booting…

Model

Fish Speech

Fish Speech is an open-weight AI model, released in 2023, made for developers.

An open-weight multilingual text-to-speech model family with voice cloning and emotion control.

Main jobs: Text to speech, Voice cloning. It is open source and free to use. You can use it offline on your own machine, on your own server, in the terminal, through an API and in the browser.

Visit speech.fish.audio

Markdown version of https://godooo.ai/en/model/fish-speech. Every page on this site has one: add .md to its address.

At a glance

Type
Model
Released
2023
Open source
Yes
GitHub stars
32,830
How you run it
Runs offline · Self-host · Terminal · API · Web
Made for
Developers
Also known as
OpenAudio · Fish Audio

What it can do

  • Text to speechmain jobunconfirmed

    “Best text-to-speech system among both open source and closed source”— speech.fish.audio, 2026-09-24
  • Voice cloningmain jobunconfirmed

    “Rapid Voice Cloning Fish Audio S2 supports accurate voice cloning using short reference samples (typically 10-30 seconds).”— github.com, 2026-09-24

unconfirmedproposed by a machine, awaiting calibration

Alternatives

Connections

Questions

What is Fish Speech?

Fish Speech is an open-weight AI model, released in 2023, made for developers. An open-weight multilingual text-to-speech model family with voice cloning and emotion control.

Is Fish Speech free?

Yes. Fish Speech is open source and free to use.

Is Fish Speech open source?

Yes. The source code is at https://github.com/fishaudio/fish-speech.

Can Fish Speech run locally?

Yes. Fish Speech can run offline on your own machine. It can also be self-hosted on your own server.

What can Fish Speech do?

Main jobs: Text to speech, Voice cloning.

What are open-source alternatives to Fish Speech?

Open-source ones: Chatterbox, CosyVoice, F5-TTS, GPT-SoVITS, ChatTTS, RVC and MoneyPrinterTurbo. Others: ElevenLabs, CapCut 剪映, Descript, WaveNet and HeyGen.

Sources