Examples

Runnable bitHuman projects grouped by what you're building — every one is open-source. Clone, set your API secret, run.

Pick your path

Every project below is open-source under bithuman-product/bithuman-sdk-public/Examples. Each tutorial page follows the same shape: Prerequisites → Run it → What you’ll see → Full code → Next steps. Find your row and start there.

If you want to…	Start with	Language	Time
Try a talking avatar end-to-end, no code	CLI — Hello, avatar	CLI	~2 min
The smallest streaming loop in code	Python — Hello, avatar	Python	~5 min
Call the platform from any language	REST — Hello, avatar	curl	~5 min
A talking voice assistant on a Mac/iPad/iPhone	Swift / iOS — Hello, avatar	SwiftUI	~15 min
An avatar on an Android phone or tablet	Kotlin / Android — Hello, avatar	Kotlin	~15 min
Full voice conversation, mic in / avatar out	AI voice chat	Python	~10 min

Note New here? The fastest end-to-end demo is the CLI — Hello, avatar: one brew install, one command, a talking avatar in your browser. No code.

No-code & smallest scripts

• CLI — Hello, avatar — install, bithuman doctor, bithuman run. Full demo, zero code.
• quickstart project — the smallest scripted path: API key, a model, your first render. Auto-downloads a sample avatar on first run.

Backend & voice agents — Python

The streaming runtime and LiveKit voice agents. Each repo project ships an .env.example, requirements.txt, and a docker compose stack.

• Python — Hello, avatar — the minimal AsyncBithuman streaming loop, ~20 lines.
• AI voice chat — OpenAI Realtime voice in, lip-synced avatar out. No server.
• python/local-essence — Essence on your own CPU box. Ships quickstart.py, microphone.py, conversation.py, plus a web UI at http://localhost:4202.
• python/cloud-essence — Essence on bitHuman cloud + LiveKit + browser UI. Start here for production agents.

Native apps — Swift & Kotlin

• Swift / iOS — Hello, avatar — SwiftUI bitHumanKit voice agent with a lip-synced avatar.
• Kotlin / Android — Hello, avatar — Maven Central AAR + Avatar.load on arm64-v8a.
• swift/ios-avatar — complete runnable SwiftUI iOS reference app (hardware gate + entitlements).
• swift/macos-voice — voice-only on-device agent: no avatar, no API key, fully offline.

Note Honesty about Swift examples. The runnable, current ones are ios-avatar and macos-voice. The sibling macos-avatar and essence-playback examples target SDK surfaces that have not yet shipped and won’t build against the current 0.8.x XCFramework — treat them as previews. A Flutter app exists in the bithuman-apps repo as a reference app, not a published code SDK.

Web & other languages

• REST — Hello, avatar — zero-to-avatar with nothing but curl. Backends, CI, any non-SDK language.
• rest-api — curl and Python scripts for every REST endpoint.
• integrations/nextjs-ui — a polished Next.js video-chat UI over LiveKit.
• integrations/gradio-web — talk to an avatar in the browser via Gradio + FastRTC.
• integrations/java-websocket — stream audio to an avatar server from Java over WebSocket.
• integrations/offline-mac — fully offline macOS integration.

For the REST contract see the API reference. For deployment shapes (LiveKit cloud, self-hosted GPU, embed widget) see the Guides.