Skip to content

User Guide

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

User Guide

Table of Contents

  1. Installation
  2. Creating a Project
  3. Managing Singers
  4. Editing Fragments
  5. Synthesis & Playback
  6. Exporting
  7. Settings

Installation

Windows

Download the latest installer from the download page and run the setup executable.

macOS / Linux

SXSEditor supports macOS and Linux via Electron Forge. Please build from source — see the Developer Guide.

Creating a Project

  1. Launch SXSEditor.
  2. Set your project BPM and time signature in the toolbar.
  3. Add a singer by clicking the + button in the singer panel.
  4. Choose to create a new singer or open an existing .sxssinger file.

Managing Singers

Creating a Singer

  1. Open the Singer Creator window.
  2. Enter a singer name and choose a color.
  3. Load a reference audio file (WAV) — this is the voice the model will mimic.
  4. Optionally add an avatar image.
  5. Run Audio Preprocessing to extract F0 and note data from the reference audio.
  6. Save the singer as a .sxssinger file.

Singer File Format (.sxssinger)

Singer files contain:

  • Singer metadata (name, color, avatar)
  • Reference audio features (F0, mel-spectrogram)
  • Preprocessing configuration

Editing Fragments

Adding a Fragment

  1. Click the + button on a singer row to add a new fragment.
  2. Double-click a fragment to open the Fragment Editor.

Piano Roll Editor

In the fragment editor you can:

  • Add notes: Click on the piano roll grid to create notes
  • Edit note properties: Adjust pitch, duration, and timing
  • Input lyrics: Enter lyrics for each note
    • Chinese: Supports both Pinyin (e.g., ni hao) and Chinese characters (e.g., 你好)
    • English: Standard English lyrics
  • Draw pitch curves: Use the pitch envelope editor for expressive control
  • Adjust envelopes: Fine-tune volume and other parameters

Saving

Save the fragment to return to the main timeline.

Synthesis & Playback

  1. Press the ▶ Play button to synthesize and play the entire project.
  2. The editor will automatically initialize the SVS pipeline and generate audio.
  3. Use ⏸ Pause and ⏹ Stop to control playback.

Tips

  • First-time synthesis may take a moment as the model loads
  • Subsequent playbacks will be faster
  • GPU acceleration will be used automatically if available

Exporting

  1. Click the 📤 Export button.
  2. The project will be synthesized and mixed.
  3. Choose a location to save the final WAV file.

Audio Specifications:

  • Sample Rate: 24000 Hz
  • Format: WAV (16-bit PCM)

Settings

Open Settings from the menu to configure:

Setting Description
Inference Device Select a specific DirectML GPU or use automatic selection

Device settings take effect after restarting the pipeline.

Clone this wiki locally