Inworld AI website screenshot

Inworld AI

Inworld AI is a real-time voice AI infrastructure provider. The Inworld platform delivers text-to-speech, speech-to-text, an end-to-end speech-to-speech Realtime API, and an OpenAI- and Anthropic-compatible LLM Router behind one API surface and one billing relationship. Inworld's voice models lead the Artificial Analysis Speech Arena and are used to power voice agents, language-learning apps, AI companions, avatar experiences, game NPCs, and Twilio-backed phone agents. The platform supports instant and professional voice cloning, voice design from natural language, lipsync-grade phoneme alignment, on-premise TTS deployment, and zero-data-retention configurations for regulated workloads.

6 APIs 24 Features
AIArtificial IntelligenceVoiceText To SpeechSpeech To TextRealtimeLLM RoutingVoice CloningConversational AIGame AI

APIs

Inworld TTS API

Inworld TTS — real-time text-to-speech API with the #1-ranked voice models on the Artificial Analysis Speech Arena. Supports the Realtime TTS-2 model (100+ languages, natural-la...

Inworld Voice API

Inworld Voice API — manage custom voices used by the TTS and Realtime APIs. Clone voices from short audio samples (instant voice cloning) or design voices from natural-language ...

Inworld STT API

Inworld STT — speech-to-text transcription API with synchronous transcribe and a streaming WebSocket endpoint. Multi-provider routing (currently Whisper variants via Groq) with ...

Inworld Realtime API

Inworld Realtime — end-to-end speech-to-speech voice pipeline (STT + LLM + TTS) exposed over WebSocket and WebRTC. OpenAI-Realtime-API-compatible event protocol (session.update,...

Inworld LLM Router API

Inworld LLM Router — OpenAI-and-Anthropic-compatible chat-completions endpoint that routes prompts across hundreds of provider models (OpenAI, Anthropic, Google, Meta, Mistral, ...

Inworld Models API

Inworld Models API — list every model available across the Router (third-party LLMs) and Inworld first-party TTS, STT, and Realtime endpoints. Returns provider, model id, capabi...

Collections

Inworld Models API

OPEN

Inworld Realtime API

OPEN

Inworld STT API

OPEN

Inworld TTS API

OPEN

Inworld Voice API

OPEN

Pricing Plans

Inworld Ai Plans Pricing

5 plans

PLANS

Rate Limits

Inworld Ai Rate Limits

4 limits

RATE LIMITS

FinOps

Inworld Ai Finops

FINOPS

Features

Realtime TTS-2 voice model — 100+ languages, natural-language steering, sub-200ms first-token latency
Realtime TTS 1.5 Max —
Realtime TTS 1.5 Mini — cost-optimized voice with ~120ms first-token latency
Instant voice cloning from short audio samples
Professional voice cloning with audio processing controls
Voice design from natural-language descriptions plus optional reference audio
Word-, character-, and phoneme-level alignment (visemes) for lipsync and avatar rendering
Custom pronunciation, pause controls, voice tags, and long-text streaming synthesis
WebSocket TTS for bidirectional streaming synthesis
Speech-to-Text via multi-provider routing (Whisper variants on Groq) with 99+ languages, prompt biasing, word timestamps, and configurable end-of-turn detection
Realtime API — speech-to-speech pipeline over WebSocket and WebRTC, OpenAI-Realtime compatible
Twilio media-stream integration for inbound and outbound phone calls
MCP server tunneling inside Realtime sessions
JWT-based realtime authentication (separate Realtime-only API keys)
LLM Router — OpenAI-and-Anthropic-compatible chat-completions over hundreds of provider models
Named reusable routers with conditional routing, A/B traffic splitting, and provider routing
Prompt caching, prompt compression, and integrated web search inside the Router
Claude-Code-compatible mode for drop-in Claude Code substitution
Zero Data Retention (ZDR) option for TTS and Realtime
On-premise TTS deployment for regulated and air-gapped environments
ElevenLabs voice-migration tool for batch-importing voice clones
Open-source Python TTS model in the inworld-ai/tts repository
Integrations with LiveKit Agents, Pipecat, LangChain, and HeyGen avatars
Unity-side runtime templates for game and avatar use cases

Event Specifications

Inworld AI Runtime WebSocket APIs

AsyncAPI description of Inworld AI's publicly documented runtime WebSocket surface. Inworld exposes three independent WebSocket endpoints: * **TTS streaming** — bidirectional te...

ASYNCAPI

Semantic Vocabularies

Inworld Ai Context

0 classes · 8 properties

JSON-LD

API Governance Rules

Inworld AI API Rules

7 rules · 2 errors 5 warnings

SPECTRAL

Example Payloads

Inworld Router Chat Completion Example

2 fields

EXAMPLE

Inworld Stt Transcribe Example

2 fields

EXAMPLE

Inworld Tts Synthesize Speech Example

2 fields

EXAMPLE

Resources

🌐
Portal
Portal
🔗
Documentation
Documentation
🚀
GettingStarted
GettingStarted
🔗
Documentation
Documentation
🔗
Documentation
Documentation
🔗
Documentation
Documentation
📝
Signup
Signup
🔑
Authentication
Authentication
🔗
Sandbox
Sandbox
🟢
StatusPage
StatusPage
👥
GitHubOrganization
GitHubOrganization
💻
SourceCode
SourceCode
💻
CodeExamples
CodeExamples
💻
CodeExamples
CodeExamples
💻
CodeExamples
CodeExamples
💻
CodeExamples
CodeExamples
💻
CodeExamples
CodeExamples
📦
SDKs
SDKs
📦
SDKs
SDKs
📦
SDKs
SDKs
💻
CodeExamples
CodeExamples
🔧
Tools
Tools
🔧
Tools
Tools
💻
CodeExamples
CodeExamples
💻
CodeExamples
CodeExamples
💻
CodeExamples
CodeExamples
💻
CodeExamples
CodeExamples
💻
CodeExamples
CodeExamples
💻
CodeExamples
CodeExamples
💻
CodeExamples
CodeExamples
💻
CodeExamples
CodeExamples
💻
CodeExamples
CodeExamples
💻
CodeExamples
CodeExamples
🔗
RateLimits
RateLimits
💰
Pricing
Pricing
💰
Pricing
Pricing
💰
Pricing
Pricing
💰
Pricing
Pricing
💰
Pricing
Pricing
🔗
Documentation
Documentation
🔗
Security
Security
🔗
Deployment
Deployment
📄
ChangeLog
ChangeLog
🔗
Migration
Migration
🔗
Migration
Migration
🔗
Migration
Migration
💬
Support
Support
💰
Pricing
Pricing
🔗
Plans
Plans
🔗
Plans
Plans
🔗
RateLimits
RateLimits
🔗
FinOps
FinOps

Sources

Raw ↑ 
opencollection: 1.0.0
info:
  name: Inworld Voice API
  version: v1
request:
  auth:
    type: basic
    username: '{{username}}'
    password: '{{password}}'
items:
- info:
    name: Voices
    type: folder
  items:
  - info:
      name: List Voices
      type: http
    http:
      method: GET
      url: https://api.inworld.ai/voices/v1/voices
      params:
      - name: pageSize
        value: ''
        type: query
      - name: pageToken
        value: ''
        type: query
    docs: List all voices in the caller's workspace.
  - info:
      name: Clone Voice
      type: http
    http:
      method: POST
      url: https://api.inworld.ai/voices/v1/voices:clone
      body:
        type: json
        data: '{}'
    docs: 'Instant voice cloning. Submit one or more short audio samples (WAV or MP3, base64-encoded) and receive a new custom
      voice that can be used immediately in TTS and Realtime calls.

      '
  - info:
      name: Design Voice
      type: http
    http:
      method: POST
      url: https://api.inworld.ai/voices/v1/voices:design
      body:
        type: json
        data: '{}'
    docs: 'Generate a brand-new voice from a natural-language description plus optional reference audio. Returns a candidate
      voice that can be previewed and then published.

      '
  - info:
      name: Publish Voice
      type: http
    http:
      method: POST
      url: https://api.inworld.ai/voices/v1/voices:publish
      body:
        type: json
        data: '{}'
    docs: Publish a designed or cloned voice so it is available for use in TTS and Realtime.
  - info:
      name: Get Voice
      type: http
    http:
      method: GET
      url: https://api.inworld.ai/voices/v1/voices/:voiceId
      params:
      - name: voiceId
        value: ''
        type: path
    docs: Get Voice
  - info:
      name: Update Voice
      type: http
    http:
      method: PATCH
      url: https://api.inworld.ai/voices/v1/voices/:voiceId
      params:
      - name: voiceId
        value: ''
        type: path
      body:
        type: json
        data: '{}'
    docs: Update Voice
  - info:
      name: Delete Voice
      type: http
    http:
      method: DELETE
      url: https://api.inworld.ai/voices/v1/voices/:voiceId
      params:
      - name: voiceId
        value: ''
        type: path
    docs: Delete Voice
bundled: true