Four models, one route
S2.1 Pro, S2 Pro and S1 for text-to-speech, plus Transcribe-1 for speech-to-text with word-level timestamps, all addressed as fish-audio/ model strings.
Fish Audio on Vercel AI Gateway
Fish Audio's speech and transcription models are available on Vercel AI Gateway, so AI SDK and Next.js teams can add expressive voice with the same key, billing and observability they already use for their LLMs.
CategoryRouters & gateways
IntegrationFish Audio models routed through Vercel AI Gateway
ModelsS2.1 Pro, S2 Pro, S1 (speech) and Transcribe-1 (transcription)
SDKAI SDK generateSpeech and transcribe
Available sinceAugust 19, 2026
Launch announcement ↗Why Vercel and Fish Audio
AI Gateway gives developers one endpoint for every model they use. Putting Fish Audio behind it means voice is one import away in any AI SDK app, without wiring up another provider.
Native in the AI SDK
Call generateSpeech or transcribe from the ai package with a Fish Audio model ID. No provider SDK, no custom client, no extra dependency.
Gateway billing and observability
Usage, spend and latency for Fish Audio show up alongside every other model in the AI Gateway dashboard.
How it works
3 steps from a Fish Audio voice to a shipping product on Vercel.
STEP 01
Enable AI Gateway
Create an AI Gateway API key in the Vercel dashboard, or rely on Vercel OIDC when your app is deployed on Vercel.
Typescript
12345678910
import {
experimental_generateSpeech as generateSpeech
} from 'ai';
const result = await generateSpeech({
model: 'fish-audio/s2.1-pro',
text: 'How are you doing today?',
});awaitwriteFile('speech.mp3',
result.audio.uint8Array);
STEP 02
Pick a Fish Audio model
Use fish-audio/s2.1-pro for production voice, s2-pro for inline tag control, s1 for emotion markers, or transcribe-1 for speech-to-text.
STEP 03
Generate speech in the AI SDK
Call generateSpeech with your text, then stream or save the audio. Spend tracking is on from the first request.
What teams build with Fish Audio and Vercel
Voice features in Next.js apps
Read-aloud, narrated summaries and voice replies added to an existing AI SDK app in an afternoon.
Transcribe, reason, speak
Transcribe-1 in, an LLM in the middle, S2.1 Pro out, all through one gateway and one bill.
Prototype to production
Start on gateway credits, keep the same code as usage grows, and watch spend per model in one place.
More partners
Open-source WebRTC stack and Agents framework. Fish Audio TTS ships as a first-party plugin and through LiveKit Inference.
Real-time infrastructure
AI video platform. Fish Audio powers the Fish voice engine inside HeyGen's AI Studio, with accent and locale controls for English. Available in both HeyGen Video Creator and LiveAvatar.
Video & avatars
Voice agent platform for phone and call centers. Import any Fish Audio voice to power your Retell agent.
Voice agent platforms
Start building with Fish Audio and Vercel
Browse Fish Audio models on AI Gateway, or talk to our partnerships team about co-marketing with Vercel.