Skip to content

Install ‐ Manual Installation Guide

Jump to bottom
erew123 edited this page Nov 26, 2024 · 15 revisions

This guide provides detailed manual installation instructions for AllTalk TTS version 2 on Windows, Linux, and Mac (theoretical) platforms.

⚡ It is preferable to use Atsetup.bat or Atsetup.sh as detailed in the Quick Setup Guides ⚡
⚡ as this will perform all the below steps automatically & build a correct Python environment. ⚡

⚡ I do not test AllTalk in custom built Python environments and due to the complexity of different ⚡
⚡ Python environments I cannot support them to any great extent, as there are far too many moving parts. ⚡

🚨 AKA, I will not support custom Python environments. The information below this point is an "if you know 🚨
🚨 what you are doing, then you can, technically speaking, install AllTalk into a custom Python environment" 🚨

Mac Support Disclaimer
AllTalk TTS support for Mac systems is currently untested & therefore theoretical. Please be aware of the following limitations:

  • The installation process and functionality on Mac systems have not been verified due to lack of testing hardware.
  • Compatibility of various TTS engines with Mac is uncertain and may vary.
  • Features requiring GPU acceleration (such as DeepSpeed and CUDA) will not function on Mac systems.
  • Some TTS engines may have limited or no support for Mac due to their development focus on other platforms.

These limitations are inherent to the TTS engines themselves and are beyond the control of the AllTalk TTS project. Improvements in Mac support would require updates from the individual TTS engine developers.

Users attempting to install and run AllTalk TTS on Mac systems should proceed with caution and be prepared for potential compatibility issues or reduced functionality.

Python & Pytorch

This is currently targeted at Python 3.11.x and PyTorch 2.2.1 & may well error or fail on anything other than these versions.

Prerequisites

Windows Prerequisites
Before installing AllTalk, ensure you have the following:

Linux Prerequisites
Depending on your Linux distribution, you'll need to install some packages for AllTalk and its TTS engines to function correctly:

  • Debian-based systems (e.g., Ubuntu): 
    sudo apt install libaio-dev espeak-ng ffmpeg gcc g++ git curl
  • RPM-based systems (e.g., CentOS, Fedora): 
    sudo yum install libaio-devel espeak-ng ffmpeg gcc g++ git curl



Mac Prerequisites (Theoretical)
For Mac systems, you'll need to install some packages using Homebrew. If you don't have Homebrew installed, you can install it by following the instructions on the [official Homebrew website](https://brew.sh/).

Once Homebrew is installed, run the following commands:

brew install espeak ffmpeg gcc git curl

You'll also need to install Xcode Command Line Tools if you haven't already:

xcode-select --install



Important: Avoid using dashes or spaces in your folder path (e.g., avoid my folder-is-this/alltalk_tts-main) as this causes issues with Python.

Manual Installation Instructions

Windows Manual Installation

  1. Open Command Prompt and navigate to your preferred directory:

    cd /d C:\path\to\your\preferred\directory

  2. Clone the AllTalk repository:

    git clone -b alltalkbeta https://github.com/erew123/alltalk_tts

  3. Navigate to the AllTalk directory:

    cd alltalk_tts

  4. Create and navigate to the alltalk_environment directory:

    mkdir alltalk_environment
cd alltalk_environment

  5. Download and install Miniconda:

    curl -Lk "https://repo.anaconda.com/miniconda/Miniconda3-py311_24.4.0-0-Windows-x86_64.exe" > miniconda_installer.exe
start /wait "" miniconda_installer.exe /InstallationType=JustMe /NoShortcuts=1 /AddToPath=0 /RegisterPython=0 /NoRegistry=1 /S /D=%cd%\conda

  6. Navigate to the conda folder and create a new environment:

    cd conda
.\condabin\conda create --no-shortcuts -y -k --prefix ..\env python=3.11.9

  7. Activate the new environment:

    call .\condabin\conda.bat activate ..\env

  8. Install PyTorch 2.2.1:

    .\Scripts\conda install -y pytorch==2.2.1 torchvision==0.17.1 torchaudio==2.2.1 pytorch-cuda=12.1 -c pytorch -c nvidia

  9. Install Faiss and FFmpeg:

    .\Scripts\conda install -y pytorch::faiss-cpu
.\Scripts\conda install -y conda-forge::ffmpeg=*=*gpl*
.\Scripts\conda install -y -c conda-forge "ffmpeg=*=h*_*" --no-deps

  10. Navigate back to the alltalk_tts folder and install requirements:

    cd ..
cd ..
pip install -r system\requirements\requirements_standalone.txt

  11. Update Gradio:

    pip install --upgrade gradio==4.32.2

  12. Download and install DeepSpeed:

    curl -LO https://github.com/erew123/alltalk_tts/releases/download/DeepSpeed-14.0/deepspeed-0.14.0+ce78a63-cp311-cp311-win_amd64.whl
pip install deepspeed-0.14.0+ce78a63-cp311-cp311-win_amd64.whl
del deepspeed-0.14.0+ce78a63-cp311-cp311-win_amd64.whl

  13. Install Parler:

    pip install -r system\requirements\requirements_parler.txt

  14. Clean the conda environment:

    .\alltalk_environment\conda\Scripts\conda clean --all --force-pkgs-dirs -y

  15. (Optional) Downgrade transformers for XTTS streaming support:

    pip install transformers==4.42.4



Linux Manual Installation

  1. Open a terminal and navigate to your preferred directory:

    cd /path/to/your/preferred/directory

  2. Clone the AllTalk repository:

    git clone -b alltalkbeta https://github.com/erew123/alltalk_tts

  3. Navigate to the AllTalk directory:

    cd alltalk_tts

  4. Create and navigate to the alltalk_environment directory:

    mkdir alltalk_environment
cd alltalk_environment

  5. Download and install Miniconda:

    curl -LO https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh
bash Miniconda3-latest-Linux-x86_64.sh -b -p $PWD/conda

  6. Create a new conda environment:

    ./conda/bin/conda create --no-shortcuts -y -k --prefix ../env python=3.11.9

  7. Activate the new environment:

    source ./conda/bin/activate ../env

  8. Install PyTorch 2.2.1:

    conda install -y pytorch==2.2.1 torchvision==0.17.1 torchaudio==2.2.1 pytorch-cuda=12.1 -c pytorch -c nvidia

  9. Install Faiss and FFmpeg:

    conda install -y pytorch::faiss-cpu
conda install -y -c conda-forge "ffmpeg=*=*gpl*"
conda install -y -c conda-forge "ffmpeg=*=h*_*" --no-deps

  10. Navigate back to the alltalk_tts folder and install requirements:

    cd ..
pip install -r system/requirements/requirements_standalone.txt

  11. Update Gradio:

    pip install --upgrade gradio==4.32.2

  12. Install DeepSpeed:

    pip install deepspeed

  13. Install Parler:

    pip install -r system/requirements/requirements_parler.txt

  14. Clean the conda environment:

    ./alltalk_environment/conda/bin/conda clean --all --force-pkgs-dirs -y

  15. (Optional) Downgrade transformers for XTTS streaming support:

    pip install transformers==4.42.4



Mac Manual Installation (Theoretical)

Note: This Mac installation process is theoretical and has not been tested. Use with caution.

  1. Open Terminal and navigate to your preferred directory:

    cd /path/to/your/preferred/directory

  2. Clone the AllTalk repository:

    git clone -b alltalkbeta https://github.com/erew123/alltalk_tts

  3. Navigate to the AllTalk directory:

    cd alltalk_tts

  4. Create and navigate to the alltalk_environment directory:

    mkdir alltalk_environment
cd alltalk_environment

  5. Download and install Miniconda:

    curl -LO https://repo.anaconda.com/miniconda/Miniconda3-latest-MacOSX-x86_64.sh
bash Miniconda3-latest-MacOSX-x86_64.sh -b -p $PWD/conda

  6. Create a new conda environment:

    ./conda/bin/conda create --no-shortcuts -y -k --prefix ../env python=3.11.9

  7. Activate the new environment:

    source ./conda/bin/activate ../env

  8. Install PyTorch 2.2.1 (CPU version, as CUDA is not available on Mac):

    conda install -y pytorch==2.2.1 torchvision==0.17.1 torchaudio==2.2.1 -c pytorch

  9. Install Faiss and FFmpeg:

    conda install -y pytorch::faiss-cpu
conda install -y -c conda-forge "ffmpeg=*=*gpl*"

  10. Navigate back to the alltalk_tts folder and install requirements:

    cd ..
pip install -r system/requirements/requirements_standalone.txt

  11. Update Gradio:

    pip install --upgrade gradio==4.32.2

  12. Install Parler:

    pip install -r system/requirements/requirements_parler.txt

  13. Clean the conda environment:

    ./alltalk_environment/conda/bin/conda clean --all --force-pkgs-dirs -y

  14. (Optional) Downgrade transformers for XTTS streaming support:

    pip install transformers==4.42.4

Note: DeepSpeed is not installed on Mac as it's primarily for NVIDIA GPUs. Some features requiring GPU acceleration may not work on Mac systems.



Additional Notes

  • DeepSpeed is automatically installed but will only work on Nvidia GPUs (Windows and Linux only).
  • Always activate the appropriate Python environment before making adjustments or using features like Fine-tuning.
  • For more detailed instructions on Fine-tuning and DeepSpeed, refer to the setup utility and the full documentation.
  • If you're new to Python environments, consider reviewing the Understanding Python Environments Simplified section in the Help documentation.
  • Mac users may experience limited functionality due to lack of GPU acceleration for some TTS engines.

AllTalk Version 2 Index

Installation

System Requirements

Features
3rd Party Integrations
XTTS Finetuning Guides
API Documentation
Support & Help
Clone this wiki locally