Skip to content

FAQ

Jump to bottom
SXSEditor Dev edited this page May 10, 2026 · 1 revision

Frequently Asked Questions (FAQ)

General

What is SXSEditor?

SXSEditor is an open-source desktop singing voice synthesis (SVS) workstation. It allows you to create, edit, and synthesize vocal tracks using a visual piano-roll editor and AI-powered neural singing voice synthesis.

Is SXSEditor free?

Yes! SXSEditor is completely free and open-source under the MIT License.

What platforms are supported?

  • Windows: Primary target with pre-built installers
  • macOS: Supported via Electron Forge (build from source)
  • Linux: Supported via Electron Forge (build from source)

Models

What models do I need?

You need the SoulX-Singer ONNX model files placed in the onnx_models/ directory. These include encoders, the diffusion model, and the vocoder.

Where do I get the models?

The SoulX-Singer model files need to be obtained separately. Check the repository documentation for download sources.

Can I use my own models?

Currently SXSEditor is designed to work with the SoulX-Singer architecture. Custom model support is not available at this time.

Synthesis

What languages are supported for synthesis?

Language Status
English ✅ Supported
Chinese (Mandarin) ✅ Supported — accepts Pinyin or Chinese characters
Japanese 🔄 In Development
Korean 📋 Planned

How does Chinese lyric input work?

You can input lyrics in either Pinyin (e.g., ni hao) or Chinese characters (e.g., 你好). The system will automatically convert characters to phonemes for synthesis.

Can I use my own voice?

Yes! Use the Singer Creator to create a custom singer from a reference WAV audio file. The model will analyze the vocal characteristics and use them during synthesis.

Hardware

Does it work without a GPU?

Yes. ONNX Runtime will automatically fall back to CPU if no compatible GPU is detected. GPU acceleration is beneficial but not required.

Which GPUs are supported?

DirectML supports NVIDIA, AMD, and Intel discrete GPUs. The application will automatically detect and select the best available device.

What are the minimum system requirements?

  • OS: Windows 10/11 (64-bit)
  • RAM: 8 GB minimum (16 GB recommended)
  • Storage: 2 GB for the application, additional space for models
  • CPU: Any modern multi-core processor

Troubleshooting

The app crashes on startup

Try the following:

  1. Ensure all model files are correctly placed in onnx_models/
  2. Check that Node.js dependencies are properly installed
  3. Look for error logs in the console
  4. Open an issue on GitHub

Synthesis is very slow

  • Enable GPU acceleration if available
  • Check that your GPU drivers are up to date
  • For CPU inference, a modern multi-core processor is recommended

Native module build errors

If you encounter errors when building from source:

npx electron-rebuild

Contributing

How can I contribute?

See the Developer Guide for detailed contribution guidelines. We welcome bug reports, feature requests, and pull requests.

I found a bug

Please open an issue on the GitHub Issue Tracker with:

  • A clear description of the problem
  • Steps to reproduce
  • Your system configuration
  • Any relevant logs or screenshots

Support

Where can I get help?

  • Open an issue on GitHub Issues
  • Check the User Guide for detailed instructions
  • Consult this FAQ for common questions

Clone this wiki locally