Fish Audio
AI text-to-speech, voice cloning, and speech-to-text with emotion control
Linked from fish.audio · checked 29 Oct
- Tool type
- AI text-to-speech and voice cloning platform
- Made for
- Creators, developers, and businesses
- Runs on
- Web app and hosted API
- Use the browser app at fish.audio or call the hosted API at api.fish.audio. Self-hosting is an Enterprise engagement.
From @FishAudio
Recent important posts · features
- New feature
Drama 3 upgraded weights rank #1 open-weight voice
“Drama 3 just leveled up 🎭 Upgraded weights. On @ArtificialAnlys: → Provider Voice: #1 open-weight, #6 overall. +38 Elo over Preview. → Controlled Voice: #1 open-weight. +22 Elo over Preview. Try Drama 3 at full power ↓”
@FishAudio · Oct 29View
What it does
Fish Audio accepts typed scripts with inline direction tags such as [emphasis] or [whispering] and a chosen voice, then produces spoken audio files such as MP3. It also transcribes speech to text, including multiple speakers and emotion tags. Voices can be selected from a large public library or cloned from a short audio sample.
Creators use it for video voiceovers, audiobook narration, character voices for games and animation, and conversational voice agents. Inline emotion and non-speech tags such as [laughing] or [long pause] give direct control over delivery. The site also offers avatar lipsync that pairs Fish Audio voices with images or video generated by third-party models.
It suits content creators producing narration and developers building voice features through a REST API with Python and TypeScript SDKs. The free web plan is limited to personal, non-commercial use, and commercial use requires a paid plan. API usage is billed separately on pay-as-you-go terms, and the S2.1 Pro Free API model is free through November 30, 2026, without an SLA.
What it can do
- Text to speech with inline emotion and non-speech direction tags
- Voice cloning from short audio samples
- Speech-to-text transcription with multispeaker and emotion tags
- Large public voice library
- Real-time streaming via WebSocket and REST API
- Python and TypeScript SDKs for API integration
Before you choose
- The free plan is for personal, non-commercial use; commercial use requires a paid plan.
- Monthly web credits reset each billing cycle, and unused minutes do not roll over.
- API usage is billed separately from web subscriptions, at $15 per million UTF-8 bytes for TTS.
Listed under
Get started
Create a free web account for a monthly credit allowance. API keys are issued from the developer section.
- Create a free account at fish.audio/auth to use the web app.
- Choose a voice and enter text with optional emotion tags on the text-to-speech page.
- For API use, obtain an API key from the developers page, then call the /v1/tts endpoint or install the Python SDK.
Plans and prices
Prices in USDPlus and Pro show annual totals ($132 and $900 billed annually). API TTS is $15 per 1M UTF-8 bytes; the S2.1 Pro Free model is free through November 30, 2026, with no SLA.
- 8,000 credits monthly
- Up to 500 characters per generation
- 250,000 credits monthly
- Up to 200 minutes generation
- 2,000,000 credits monthly
- 3 team seats included
Source: fish.audio/plan, checked 29 Oct 2026. Prices can change; check before you buy.
Sources for this overview
Product pages, documentation, and pricing checked 29 Oct 2026. Each source supports part of the overview; availability and prices can change.
- Best AI Text To Speech & Free Voice Cloning | Fish Audio ↗fish.audio
- Pricing & Plans - Fish Audio ↗fish.audio
- Developers - Fish Audio ↗fish.audio
Ratings and reviews
No ratings for Fish Audio yet.
What’s changed
Free accountSignals
Free accountData and security
Free accountCompare with similar tools
Free accountWhere this comes from
How we research- Company website
- fish.audio: product, pricing and policy pages
- YC directory mirror
- Batch and company name
- Social accounts
- Accounts the company links, and its posts on X
- Human review
- Checked by a person
- knowledge_base
Last checked 29 Oct 2026. Every field keeps its source and date; unknown means we couldn’t confirm it.