AI VTuber Live Streaming Setup - Open-LLM-VTuber + YouTube Live + Edge TTS on AWS

🎭 AI VTuber Live Stream

Fully automated AI VTuber that livestreams on YouTube, reads chat in real-time, and responds with voice — powered by Open-LLM-VTuber.

🔴 Live Demo

Watch it live: https://youtube.com/watch?v=7V90Jx7Eygo

🏗️ Architecture

CODE

┌─────────────────────────────────────────────────────────────────────┐  
│                         AWS EC2 (Ubuntu 26.04)                      │  
│                        t3.large (2 vCPU, 8GB)                       │  
│                                                                     │  
│   ┌────────────┐     ┌─────────────────────┐     ┌──────────────┐  │  
│   │  YouTube   │────▶│  Open-LLM-VTuber    │────▶│  LLM API     │  │  
│   │  Chat      │     │  (Python Server)    │     │  (OpenAI     │  │  
│   │  Bridge    │     │  :12393 / :12394    │     │  Compatible) │  │  
│   └────────────┘     └──────────┬──────────┘     └──────────────┘  │  
│                                 │                                   │  
│                    ┌────────────┼────────────┐                      │  
│                    ▼            │            ▼                      │  
│            ┌────────────┐      │    ┌──────────────┐               │  
│            │  Edge TTS  │      │    │  Live2D Model │               │  
│            │  (Voice)   │      │    │  (Animation)  │               │  
│            └─────┬──────┘      │    └──────┬───────┘               │  
│                  │             │           │                        │  
│                  ▼             │           ▼                        │  
│          ┌────────────┐       │    ┌──────────────┐                │  
│          │ PulseAudio │       │    │   Chrome      │                │  
│          │ (Virtual   │       │    │   (Display    │                │  
│          │  Sink)     │       │    │    :99)       │                │  
│          └─────┬──────┘       │    └──────┬───────┘                │  
│                │              │           │                         │  
│                └──────────────┼───────────┘                         │  
│                               ▼                                     │  
│                       ┌──────────────┐                              │  
│                       │   FFmpeg     │                              │  
│                       │ (x11grab +   │                              │  
│                       │  pulse)      │                              │  
│                       └──────┬───────┘                              │  
│                              │                                      │  
└──────────────────────────────┼──────────────────────────────────────┘  
                               │ RTMP  
                               ▼  
                    ┌─────────────────────┐  
                    │   YouTube Live      │  
                    │   (RTMP Ingest)     │  
                    └─────────────────────┘

📦 All Dependencies (with full setup for each)

Each dependency has its own folder with README, install script, and config files:

#	Dependency	Folder	GitHub	What We Use It For
1	Open-LLM-VTuber	`dependencies/open-llm-vtuber/`	GitHub	Core VTuber framework (server, frontend, Live2D, proxy)
2	blivedm	`dependencies/blivedm/`	GitHub	Bilibili live chat integration
3	edge-tts	`dependencies/edge-tts/`	GitHub	Free Microsoft Edge text-to-speech
4	chat-downloader	`dependencies/chat-downloader/`	GitHub	YouTube live chat scraper
5	websocket-client	`dependencies/websocket-client/`	GitHub	Python WebSocket client
6	Google Chrome	`dependencies/chrome/`	Download	Live2D rendering + audio playback
7	FFmpeg	`dependencies/ffmpeg/`	ffmpeg.org	Screen + audio capture → RTMP stream
8	PulseAudio	`dependencies/pulseaudio/`	freedesktop.org	Virtual audio sink for routing
9	Xvfb	`dependencies/xvfb/`	X.Org	Virtual X11 display for headless Chrome
10	sherpa-onnx	`dependencies/sherpa-onnx/`	GitHub	Local speech recognition (ASR)

Python Packages

CODE

chat-downloader    # YouTube chat scraping  
websocket-client   # WebSocket connections  
edge-tts           # Free TTS  
loguru             # Logging  
requests           # HTTP requests  
sherpa-onnx        # Local ASR  
openai             # LLM API client  
anthropic          # Claude API client

🚀 One-Command Setup

BASH

curl -sSL https://raw.githubusercontent.com/Maliot100X/ai-vtuber-livestream/main/setup.sh | bash

Or install each dependency manually from its folder above.

📋 Prerequisites

AWS EC2 — Ubuntu 24.04+, t3.large (2 vCPU, 8GB RAM)
Security Group — Open ports 12393, 12394, 22
YouTube Channel — Live streaming enabled
LLM API Key — Any OpenAI-compatible provider

🔧 Manual Setup (Step by Step)

Step 1: System Dependencies

BASH

sudo apt update && sudo apt upgrade -y  
sudo apt install -y python3 python3-pip python3-venv python3-dev \  
  git curl wget unzip build-essential \  
  xvfb pulseaudio ffmpeg \  
  libnss3 libatk-bridge2.0-0 libdrm2 libxkbcommon0 \  
  libgbm1 libasound2 libxshmfence1 libgtk-3-0

Step 2: Install Chrome

BASH

# See: dependencies/chrome/  
wget -q -O /tmp/chrome.deb https://dl.google.com/linux/direct/google-chrome-stable_current_amd64.deb  
sudo dpkg -i /tmp/chrome.deb || sudo apt-get install -f -y

Step 3: Install Open-LLM-VTuber

BASH

# See: dependencies/open-llm-vtuber/  
cd ~  
git clone https://github.com/Open-LLM-VTuber/Open-LLM-VTuber.git  
cd Open-LLM-VTuber  
python3 -m venv .venv && source .venv/bin/activate  
pip install -r requirements.txt  
pip install chat-downloader websocket-client edge-tts loguru requests sherpa-onnx

Step 4: Configure

BASH

cp ~/ai-vtuber-livestream/config/conf.yaml.full ~/Open-LLM-VTuber/conf.yaml  
cp ~/ai-vtuber-livestream/.env.example ~/Open-LLM-VTuber/.env  
nano ~/Open-LLM-VTuber/.env  # Add your API keys

Step 5: Patch Frontend

BASH

# See: patches/frontend-proxy-ws-fix.md  
cd ~/Open-LLM-VTuber  
FRONTEND_JS=$(ls frontend/assets/main-*.js)  
sed -i 's|/client-ws|/proxy-ws|g' "$FRONTEND_JS"

Step 6: Start Everything

BASH

~/ai-vtuber-livestream/scripts/start_vtuber.sh

📁 Project Structure

CODE

ai-vtuber-livestream/  
├── README.md  
├── LICENSE  
├── .env.example  
├── .gitignore  
├── setup.sh  
│
├── config/  
│   ├── conf.yaml.example              # Minimal config template  
│   └── conf.yaml.full                 # Complete 504-line config  
│
├── scripts/  
│   ├── start_vtuber.sh                # Start all services  
│   ├── stop_vtuber.sh                 # Stop all services  
│   ├── health_check.sh                # Status check  
│   ├── vtuber-chrome.sh               # Chrome launch  
│   ├── youtube-stream.sh              # FFmpeg RTMP stream  
│   ├── yt-chat-reader.py              # Chat bridge (Method 1)  
│   ├── yt-chat-reader2.py             # Chat bridge (Method 2)  
│   ├── youtube_chat_bridge.py         # Threaded bridge  
│   ├── youtube_chat_bridge_v2.py      # Bridge v2  
│   ├── youtube_live_bridge.py         # Full bridge  
│   └── run_bilibili_live.py           # Bilibili integration  
│
├── dependencies/                      # ← FULL SETUP FOR EACH DEPENDENCY  
│   ├── open-llm-vtuber/               # Core framework  
│   │   ├── README.md  
│   │   ├── install.sh  
│   │   └── conf.yaml.working          # Our actual working config  
│   ├── blivedm/                       # Bilibili live chat  
│   │   └── README.md  
│   ├── edge-tts/                      # Free TTS  
│   │   ├── README.md  
│   │   └── install.sh  
│   ├── chat-downloader/               # YouTube chat scraper  
│   │   ├── README.md  
│   │   ├── install.sh  
│   │   ├── yt-chat-reader.py          # Our working script  
│   │   └── yt-chat-reader2.py         # Alternative method  
│   ├── websocket-client/              # WebSocket client  
│   │   ├── README.md  
│   │   └── install.sh  
│   ├── chrome/                        # Browser rendering  
│   │   ├── README.md  
│   │   ├── install.sh  
│   │   └── launch.sh                  # Our Chrome launch script  
│   ├── ffmpeg/                        # RTMP streaming  
│   │   ├── README.md  
│   │   ├── install.sh  
│   │   └── stream.sh                  # Our FFmpeg stream script  
│   ├── pulseaudio/                    # Virtual audio  
│   │   ├── README.md  
│   │   └── install.sh  
│   ├── xvfb/                          # Virtual display  
│   │   ├── README.md  
│   │   └── install.sh  
│   └── sherpa-onnx/                   # Local ASR  
│       ├── README.md  
│       └── install.sh  
│
├── patches/  
│   └── frontend-proxy-ws-fix.md       # Frontend patch instructions  
│
└── docs/  
    ├── ARCHITECTURE.md                # Deep dive  
    ├── TROUBLESHOOTING.md             # Common fixes  
    └── LLM_PROVIDERS.md               # 8+ providers

🎮 Management Commands

BASH

# Start everything  
~/ai-vtuber-livestream/scripts/start_vtuber.sh

# Stop everything  
~/ai-vtuber-livestream/scripts/stop_vtuber.sh

# Check status  
~/ai-vtuber-livestream/scripts/health_check.sh

# View logs  
tail -f /tmp/vtuber-server.log  
tail -f /tmp/yt-bridge.log  
tail -f /tmp/ffmpeg-stream.log

🤖 Supported LLM Providers

Provider	Config Key	Example Model
Groq	`groq_llm`	`llama-3.3-70b-versatile`
DeepSeek	`deepseek_llm`	`deepseek-chat`
OpenAI	`openai_llm`	`gpt-4o-mini`
OpenRouter	`openai_compatible_llm`	`anthropic/claude-3.5-sonnet`
Gemini	`gemini_llm`	`gemini-2.0-flash-exp`
Claude	`claude_llm`	`claude-3-haiku`
Ollama	`ollama_llm`	`llama3.2`

See docs/LLM_PROVIDERS.md

📄 License

MIT — see LICENSE

Built with ❤️ for the VTuber community

Source: https://github.com/Maliot100X/ai-vtuber-livestream

Agent

ai-vtuber-livestream

About this agent

🎭 AI VTuber Live Stream

🔴 Live Demo

🏗️ Architecture

📦 All Dependencies (with full setup for each)

Python Packages

🚀 One-Command Setup

📋 Prerequisites

🔧 Manual Setup (Step by Step)

Step 1: System Dependencies

Step 2: Install Chrome

Step 3: Install Open-LLM-VTuber

Step 4: Configure

Step 5: Patch Frontend

Step 6: Start Everything

📁 Project Structure

🎮 Management Commands

🤖 Supported LLM Providers

📄 License

Comments & Discussion

Tags

Share