Fish Audio
Expressive TTS and voice cloning API with open-weight models
About Fish Audio
Fish Audio offers text-to-speech, voice cloning, voice design, and speech-to-text through a web app and a pay-as-you-go API. Its S2.1-Pro model supports 83 languages with inline emotion and direction tags, streams over WebSocket, and costs $15 per million UTF-8 bytes. The fish-speech, S1, and S2 models ship as open weights under a paid commercial license, and enterprises can self-host them.
Pros & Cons
Pros
- Low, published API price with no subscription required
- Inline emotion tags and a large community voice library suit character and companion voices
- 83 languages on the current model
- Open weights give a path to self-hosting for teams that outgrow the API
Cons
- Concurrency starts at 5 requests until you prepay $100 or more
- Self-hosting the open-weight models commercially is an expensive Enterprise engagement
- Community voices vary in quality and rights clarity, so shipped products should use owned or cloned voices
Key Features
S2.1-Pro text-to-speech
Current flagship TTS model with 83 languages, automatic language detection, and inline emotion and paralinguistic cues.
Inline direction tags
Tags such as [warm], [near-whisper], [chuckle], and [long pause] travel with the text to steer delivery.
Voice cloning
Clone voices from short samples, with professional voice slots on paid creator plans.
Voice design
Generate a new voice from a description, billed at $0.01 per successful API request.
WebSocket streaming
Realtime streaming TTS and ASR with Python and TypeScript SDKs.
Community voice library
Large library of public, community-made voices usable in the app and API.
Open-weight models
fish-speech, S1, and S2 available as open weights with a paid commercial license; self-hosting is an Enterprise engagement.
MCP server
Official MCP server for using Fish Audio from coding assistants and agents.
Pricing
- No subscription or minimums
- S2.1-Pro and S2-Pro models
- 5 concurrent requests, rising to 15 at $100 paid and 50 at $1,000 paid
- WebSocket streaming
- 8,000 credits monthly
- Up to 7 minutes generation
- 3 public voice slots
- Commercial use
- 250,000 credits monthly
- Up to 200 minutes generation
- Voice design
- 1 professional voice slot
- 2,000,000 credits monthly
- Up to 1,620 minutes generation
- 3 team seats
- 5 professional voice slots
- 25,000,000 credits monthly
- Up to 6,250 minutes generation
- 10 team seats
- 15 professional voice slots
- Open-weight models in your VPC or on-prem
- Fine-tuning and data residency
- Paid commercial license
Best For
AI companions
Expressive character voices for consumer companion and roleplay apps.
AI avatars
Lip-syncable, emotion-aware TTS for avatar and video products.
Content voiceover
Narration and character voices for creators via the web app.
Similar Tools
ElevenLabs
AI voice generator and voice agents platform
Murf AI
AI voice generator with 200+ realistic text-to-speech voices
TTSOpenAI
Advanced AI voice engine for natural text-to-speech
LOVO AI
AI voice generator and video editor with 500+ voices in 100+ languages
Groq
Ultra-fast AI inference powered by custom LPU silicon
Fliki
Turn text into videos with AI voices in minutes