Installation

Installation

This guide covers installing Speakora on various platforms.

Prerequisites

• Python 3.10+ (required for SeamlessM4T)
• Git for cloning the repository
• FFmpeg for video/audio processing
• Node.js 18+ (optional, for frontend development)

GPU Support

For best performance, a GPU is recommended:

Platform	Requirements
NVIDIA	CUDA 11.8+ with 8GB+ VRAM
Apple Silicon	macOS 12.0+ with Metal/MPS
AMD	ROCm 5.0+

Quick Install

# Clone the repository
git clone https://github.com/rennerdo30/speakora.git
cd speakora

# Run the setup script
./setup.sh

# Start the application
./start.sh

The setup script will:

1. Create a Python virtual environment
2. Install all Python dependencies
3. Install frontend dependencies (if Node.js is available)
4. Build the frontend (if Node.js is available)
5. Create necessary directories

Manual Installation

1. Create Virtual Environment

python3.10 -m venv venv
source venv/bin/activate  # Linux/macOS
# or
venv\Scripts\activate  # Windows

2. Install Python Dependencies

pip install --upgrade pip setuptools wheel
pip install -r requirements.txt

3. Install Development Dependencies (Optional)

pip install -r requirements-dev.txt

4. Install Frontend (Optional)

cd frontend
npm install
npm run build
cd ..

5. Download Model

./run.sh download --model-size medium

Available model sizes:

Size	Model	VRAM	Quality
medium	SeamlessM4T v1	~8GB	Good
large	SeamlessM4T v2	~16GB	Best

Platform-Specific Notes

macOS

Install FFmpeg via Homebrew:

brew install ffmpeg

Ubuntu/Debian

sudo apt update
sudo apt install ffmpeg python3.10 python3.10-venv

Windows

1. Install Python 3.10+ from python.org
2. Install FFmpeg from ffmpeg.org
3. Add FFmpeg to your PATH

Docker Installation

For containerized deployment, see the Docker guide.

Troubleshooting

GPU Not Detected

Check device detection:

./run.sh info

Force CPU mode:

export SEAMLESS_DEVICE=cpu
./start.sh

Out of Memory

Use a smaller model:

# In config/default.yaml
model:
  size: "medium"

Model Download Issues

Manually download models:

./run.sh download --model-size medium

Next Steps

• Quick Start Guide
• Configuration Options