Skip to content

OwnAudioSharp Version 2.5.4

Choose a tag to compare

View all tags
@ModernMube ModernMube released this 11 Jan 07:36
· 658 commits to master since this release
V2.5.4
82b8a4a

Release Notes - Version 2.5.4

🎵 HTDemucs 4-Stem Audio Separation

This release introduces HTDemucs, a state-of-the-art AI model for professional-grade audio stem separation.

Key Features

  • 4-Stem Separation: Isolate vocals, drums, bass, and other instruments independently
  • Margin-Trimming Technology: Advanced processing eliminates volume fluctuations at chunk boundaries
  • Hybrid Transformer Architecture: Superior separation quality using the latest AI techniques

📦 Model Distribution

Important for Source Code Users:

Due to GitHub's file size limitations (166 MB model), the HTDemucs model must be downloaded manually from Hugging Face when building from source:

  • Download: htdemucs.onnx
  • Place in the models directory alongside other vocal remover models

NuGet Package Users: The model is included in the NuGet package - no manual download required!

💻 Example Code

A complete working example demonstrating HTDemucs usage is available in the Examples/Ownaudio.Example.HTDemucs/ directory.

🖥️ Windows ARM64 Support

Added native audio engine support for Windows ARM64 devices:

  • miniaudio native libraries for ARM64
  • PortAudio native libraries for ARM64

This enables full native audio performance on ARM-based Windows devices (e.g., Surface Pro X, Snapdragon-powered laptops).

Full Documentation: OwnAudioSharp Website

Assets 4
Loading