Fish Audio
Inspect the documented capabilities, access conditions and sources for your task.
Initial research: 2026-09-21. Check current terms before choosing.
Automated fact checks · 6 fields
AI extraction with a second evidence validation pass. These dates apply to the listed fields.
Updated
The facts
- Version
- S2.1 Pro / s2-pro
- Price
- From $15 per month. Subscription. Included usage and billing terms vary by plan.
- API
✓Supported - Open weights
✓Supported - Commercial use
- Free tier (s2.1-pro-free) may have restrictions on certain commercial scenarios; products generating more than $1M ARR should contact Fish Audio before use; paid plans available for full commercial licensing
- Voice Cloning
✓Supported S2.1 Pro supports voice cloning from reference audio via API across all 83 supported languages
Source for Voice Cloning (opens in a new tab)- Realtime Streaming
✓Supported ~90ms Time-to-First-Audio on standard API, designed for live voice agents and turn-taking dialogue
Source for Realtime Streaming (opens in a new tab)- Languages
✓Supported Supports 83 languages including English, Japanese, Chinese, Korean, Spanish, Arabic, French, German, Portuguese, Russian and more; single model, no separate endpoints
Source for Languages (opens in a new tab)- Emotion Control
?Not verified Not confirmed in fetched sources.
Source for Emotion Control (opens in a new tab)
Commercial risk
No litigation found in fetched sources, but note the framing in coverage of its $52m seed: voice cloning tech 'that can outlive any takedown' — the open weights mean cloned voices cannot be recalled once distributed. That is a reputational and compliance consideration for a studio, not a licensing one.
From the publisher
Automatic source checks. These do not change the verification date of the facts above.
Sources
Beyond feature support
Quality evidence.
How well a model performs depends on the task, version, and settings.
We have not published a reviewed quality evaluation for these versions yet. Feature support and price do not establish output quality.
Try the same small task with each option. Check accuracy, consistency, and how much editing you need to get a usable result.