What it is and who it's for
Fish Audio (OpenAudio) is expressive, actor-grade TTS with two paths: open OpenAudio S1 weights you can run locally, and a convenient cloud API and playground. It's for content creators and developers who need emotional voiceover plus the flexibility of "local or cloud". It stands out because top-tier quality is available both as open weights and as a managed service.
Key features
- OpenAudio S1 — open weights, top of the blind TTS-Arena, runs locally.
- Cloud API and web playground for a fast start with no hardware.
- Expressive, emotional delivery with style control.
- Voice cloning from a sample.
- Multilingual, Russian included.
- Streaming for interactive scenarios.
Get started in 5 minutes
- The fast path is the cloud: sign up at fish.audio, try synthesis in the browser, and grab an API key when you need one.
- The local path: install the open-source engine and pull the OpenAudio S1 weights:
pip install fish-speech
- Synthesize via the repo's CLI/Python (see the OpenAudio README) or call the cloud API with your key.
When to use it, when not to
- ✅ Use it if you want the most expressive voiceover possible and want to choose between local and cloud.
- ✅ Use it if top-tier open weights for a private deployment matter to you.
- ❌ Skip it — go with "ElevenLabs" — if you need mature production dubbing and the widest out-of-the-box language support.
- ❌ Skip it — go with "Kokoro TTS" — if your priority is lightness and minimal resources without cloning.
The honest price
Open weights are free (you pay in hardware). The cloud API and playground are usage-based with a trial allowance. Check the open-weights license for commercial use. Exact numbers are on the vendor's site and they change.
Gotchas
- Two paths (open vs cloud) mean two different setups; don't mix up the docs.
- Running locally needs a GPU and careful dependency installation.
- Verify the open-weights license terms before commercial use.
- Voice cloning carries the same ethical and legal limits as everywhere else.
🤖 Prompt accelerator
"I'm choosing between running OpenAudio S1 locally and using the Fish Audio cloud for expressive Russian voiceover. Help me decide based on my constraints <privacy / budget / whether I have a GPU>, and for the path I pick give me step-by-step setup: installation (or getting an API key), a synthesis example with emotion control and cloning from a sample, and a license check for commercial use.