tools
Fish Audio Review 2026 — Open-Source Voice Cloning and Real-Time TTS
Fish Audio delivers ElevenLabs-quality voice cloning for free. Open-source Fish Speech, real-time TTS API, Docker deployment, 8 languages. Best self-hosted voice AI 2026.
Fish Audio is a web platform and API for generating speech from text, cloning voices, designing voices from prompts, converting speech to text, changing voices, translating audio, and creating longer-form audio projects. It is used by creators, developers, publishers, educators, game teams, and voice-agent builders.
The free plan has monthly credit and generation limits. API access uses separate pay-as-you-go pricing, while commercial rights and production guarantees vary by plan and use case.
Features
- Generate speech from text with expressive voice controls
- Clone voices from reference audio
- Design custom voices from text prompts
- Convert speech to text with an API
- Stream TTS and ASR over WebSocket
- Create voiceovers, audiobooks, and longer audio projects
- Translate audio and separate voices from other sounds
- Open-weight Fish models require a paid commercial license
Use cases
- Create voiceovers for videos, ads, and explainers
- Build real-time voice agents and conversational applications
- Produce audiobooks and long-form narration
- Generate game and character dialogue
- Localize content while preserving a speaker's voice
- Prototype speech features with the Fish Audio API
Pros
Cons
Latest updates
- transcribe-1 retirement (transcribe-1)
transcribe-1 is deprecated and retires December 31, 2026; afterward, its requests are served and billed as transcribe-1-pro.
- More normalization languages
Normalization adds Turkish and Tagalog rules, Brazilian Portuguese support, and Traditional Chinese detection.
- TTS language field
TTS requests now accept an optional language field to select normalization and timestamp language.
- Drama 3 (Drama 3)
Drama 3 text-to-speech model added; direct tone, pacing, and character in plain language using model ID drama-3.
- Fish Audio S1 retirement (S1)
S1 is deprecated and retires December 31, 2026; requests specifying s1 will then use s2.1-pro at the same price.
Capabilities
- Text to speech — “Text to Speech with the most natural and human sounding AI voice generator” source
- Voice cloning — “Clone Any Voice with perfect fidelity in 15 seconds” source
- Speech to text — “Include multispeaker, emotion tags & natural language description in your transcription” source
- Sound effects — “Sound Effects” source
- Dubbing — “Audio Translation” source
- Many languages — “Speak 30+ languages with any voice” source
- Commercial use — “Commercial use allowed” source
- API — “Fish Audio offers the best AI voice generator API for developers with ultra-low latency, comprehensive SDKs, and simple REST endpoints.” source
Get it
Pricing
- Starting price
- $11/mo
- Prices checked
- 2026-11-06
Free Tier
Free
- 8,000 credits monthly
- Up to 7 minutes generation
- Up to 500 characters per generation
- 3 public voice slots
- Standard generation speed
- Enhanced voice cloning
Plus
- $11 mo
- $132 billed annually
- $15 mo
- 250,000 credits monthly
- Up to 200 minutes generation
- Up to 15,000 characters per generation
- Unlimited public + 10 private voice slots
- Priority generation on our latest models
- Access to Voice Design
Pro
- $75 mo
- $900 billed annually
- $100 mo
- 2,000,000 credits monthly
- Up to 1,620 minutes generation
- 3 team seats included
- Up to 30,000 characters per generation
- Unlimited voice slots
- 5 professional voice slots
Max
- 749 USD
- 8988 USD
- 999 USD