Docker Support (Beta)

The Docker build for AllTalk including these instructions, is currently in Beta. While functional, it is still undergoing testing and refinement. Please keep the following caveats in mind:

---

Important Notes

• Windows Compatibility: Not fully tested. Some users may encounter issues.
• CPU-Only Usage: Compatibility not guaranteed. Currently tested with GPU.
• Root User: The Python application may run as root, which is not recommended for production environments. This will be addressed in a future update.
• Gradio Integration: Gradio UI functionality has not been fully validated, but the general UI works.

For any feedback, questions, or issues, please join the discussion at the following link:
➡️ Docker Build Discussion

---

Features of the Docker Build

1. TTS Engine: Downloads and enables XTTS as the default TTS engine.
2. RVC Models: Automatically downloads and enables all supported RVC models.
3. DeepSpeed Integration: DeepSpeed is built and enabled by default for performance optimization.
4. Scripts for Simplification:
   • docker-build.sh: Automates building of the conda environment and DeepSpeed.
   • docker-start.sh: Simplifies container startup with various optional arguments.
5. Documentation: A detailed DOCKER_README.md is included in the repository, covering arguments and setup.

---

Quick Start Guide

Prerequisites

• Linux or Windows with Docker installed.
• NVIDIA GPUs (if using GPU support).
  See platform-specific setup instructions below.

Build and Run

1. Clone the repository and navigate to its root directory:

   git clone https://github.com/erew123/alltalk_tts.git
   cd alltalk_tts

2. Build the Docker image:

   ./docker-build.sh

3. Start the Docker container:

   ./docker-start.sh

4. Open the application in your browser at:
   http://localhost:7851/

---

Platform-Specific Setup

Linux

Ubuntu GPU Setup

1. Ensure the latest NVIDIA drivers are installed:

   sudo ubuntu-drivers install

2. Install Docker. Follow the official Docker documentation.
3. If GPU errors occur, install the NVIDIA Docker Container Toolkit:

   sudo nvidia-ctk runtime configure --runtime=docker
   sudo systemctl restart docker

Windows (WSL2)

1. Install WSL2 via PowerShell:

   wsl --install

2. Set up CUDA Toolkit:

   wget https://developer.download.nvidia.com/compute/cuda/repos/wsl-ubuntu/x86_64/cuda-keyring_1.1-1_all.deb
   sudo dpkg -i cuda-keyring_1.1-1_all.deb
   sudo apt-get update
   sudo apt-get -y install cuda-toolkit-12-4

3. Install Docker Desktop with WSL2 backend and restart your system.

---

Customization Options

Build Arguments

• TTS Model: Choose between piper, vits, or xtts:

  ./docker-build.sh --tts_model piper

• Clean Build: Remove existing dependencies and rebuild:

  ./docker-build.sh --clean

Start Arguments

• Custom Config: Use a specific JSON config file:

  ./docker-start.sh --config /path/to/config.json

• Disable UI: Skip Gradio UI exposure:

  ./docker-start.sh --no_ui

For a full list of options, refer to the DOCKER_README.md.

---

Known Limitations

• Docker Image Size: The image is approximately 21 GB, with additional space required for build processes.
• CUDA Support: Currently supports CUDA 12.4. Upgrading to newer CUDA versions is not yet possible due to dependency constraints.

---

Contributing and Feedback

This build is a work-in-progress, and community feedback is vital for its improvement. If you encounter any issues, have suggestions, or want to contribute, please visit:

➡️ Docker Build Discussion