this site is currently down because i cant afford maintaing my own server it has caused issues with my CPU temps (mid tier imac) sorry for the inconveneience # Professional Audio Stem Separator ## Discord community : https://discord.gg/qrt3gd8t A server-based audio processing tool that separates audio tracks into individual stems (vocals, drums, bass, etc.) using the **Demucs** machine learning engine. Ideal for legitimate audio engineering, podcast production, and forensic analysis. ## Powered by Demucs This service utilizes the [**Demucs**](https://github.com/facebookresearch/demucs) audio source separation engine developed by Facebook Research: - Licensed under [MIT License](https://github.com/facebookresearch/demucs/blob/main/LICENSE) - State-of-the-art deep learning architecture - Professional-grade separation quality - Supports 4-stem separation (vocals, drums, bass, other) ## Key Features - 🚀 **Server-Side Processing** - Heavy computations handled remotely - ⏱️ **Fast Results** - Get separated stems in ZIP format - 🎚️ **Studio Quality** - Demucs-powered separation algorithms - 💾 **No Installations** - Fully web-based (no plugins required) - 🆓 **Free to Use** - Open access for legitimate purposes ## Acceptable Use Cases - 🎙️ Podcast vocal isolation - 🔊 Audio restoration/cleanup - 🔍 Forensic audio analysis - 🎼 Music education/study - 🎧 Audio engineering workflows ## Strict Usage Policy⚠️ **WARNING: COPYRIGHT COMPLIANCE REQUIRED** This tool may only be used with **legally owned audio content**. By using this service, you agree: 1. ✅ **YOU MUST OWN** all uploaded content OR have explicit rights to modify it 2. ❌ **NEVER UPLOAD** copyrighted material without authorization 3. ❌ **DO NOT PROCESS** third-party content (music, films, TV shows) 4. ✅ **YOU ARE RESPONSIBLE** for ensuring content compliance > **Legal Note:** Separation of copyrighted material without permission constitutes copyright infringement. Violators may face permanent IP bans and legal consequences. ## How to Use 1. Access the web app:(https://vicsanity623.github.io) 2. Click "Choose File" to select an audio file (MP3/WAV/FLAC) 3. Select stem separation type (Vocals, Instruments, etc.) 4. Click "Separate Audio" to upload 5. Wait for server processing (progress bar will display) 6. Download ZIP file with separated stems when completed ## Supported Formats - Input: MP3, WAV, FLAC (≤100MB) - Output: ZIP containing WAV stems (Demucs standard format) ## Technical Architecture This server is using CPU only processing, therefore this is going to be very slow to process audio files. Plans on upgrading server for GPU is in the works. graph LR A[User Browser] --> B[Upload Audio] B --> C[Demucs Server] C --> D[Stem Separation] D --> E[ZIP Creation] E --> F[Result Download] ## Privacy & Data Handling Uploaded files are automatically deleted after processing No personal data collection No long-term storage of audio content Processing occurs in temporary memory only Limitations Maximum file size: 100MB Maximum duration: 15 minutes per track 3 simultaneous processes per IP address Server availability depends on resource load Frequently Asked Questions Q: Can I use this for commercial purposes? A: Yes, provided you own the audio rights and comply with our terms. Q: Why server-side processing? A: Demucs requires significant computational resources for professional-quality results. Q: What separation models are used? A: We use the latest Demucs v4 hybrid transformer model. ## Acknowledgments This project implements the Demucs audio separation system: @article{defossez2021hybrid, title={Hybrid Spectrogram and Waveform Source Separation}, author={Défossez, Alexandre}, journal={arXiv preprint arXiv:2111.03600}, year={2021} } Legal MIT License for the Demucs implementation Terms of Service Privacy Policy User Responsibility Notice: This service only provides separation technology. Users are solely responsible for copyright compliance and ensuring they have appropriate rights to process uploaded audio content. © 2025 Audio Stem Separator | Ethical AI Audio Processing