Easy Voice Toolkit is a user-friendly, locally deployable AI audio processing suite for voice recognition, transcription, and conversion. It targets users who need an integrated workflow for audio manipulation, from raw files to speech models, with a focus on voice conversion.

How It Works

The toolkit integrates several open-source projects, including audio-slicer, VoiceprintRecognition, whisper, and GPT-SoVITS. It provides a modular approach, allowing users to select specific tools or chain them for a complete voice conversion workflow. This architecture facilitates a gradual transformation of audio files into speech models.

Quick Start & Requirements

• Installation: Clone the repository with submodules (git clone --recurse-submodules) and install dependencies using pip install -r requirements.txt.
• Prerequisites: Python 3.8+, PyTorch (with CUDA 11.8 example provided).
• System: Currently supports Windows only.
• Resources: A "Ready-to-use portable package" is available for easier setup, containing all dependencies and models.
• Demo: Google Colab demo available.

Highlighted Details

• Supports Chinese, English, and Japanese for most functions.
• Includes tools for audio processing, voice recognition, transcription, dataset creation, model training, and voice conversion.
• Offers both lightweight installer and a large, ready-to-use portable package.
• Future features include LLM integration and a C++ (Qt) client refactor.

Maintenance & Community

• Active development indicated by "WIP" for backend development.
• Contact details provided for feedback and suggestions.

Licensing & Compatibility

• The project is free and open-source ("Natürlich~♪").
• Users are responsible for dataset authorization.
• Distribution or public sharing requires attribution to the original author and source.
• Not intended for production environments.

Limitations & Caveats

• Currently limited to Windows OS.
• Users must manage dataset authorization and are solely responsible for any infringement issues.
• Distribution terms require clear indication of voice changing usage and input source details.