Duplicate audio separation space
Extract voice from audio file
Separate audio into harmonic and percussive components
Convert audio using voice models and separate vocals
Separate different speakers in an audio conversation
Generate split audio tracks from a file
Separate audio stems and convert to MIDI
Separate vocals from audio/video files
whisperx-test
Audio-Separator by Politrees
music-transform
Separate vocals and instruments from audio
VoiceReplacer
PyTorch Music Source Separation is a powerful, open-source tool designed to separate vocals and instruments from music tracks. Built on top of the PyTorch framework, it leverages deep learning models to isolate individual audio sources within a mixed audio file. This tool is particularly useful for music producers, researchers, and developers who need high-quality audio separation for various applications.
• Real-time processing: Capable of separating audio sources in real-time for live applications.
• Pre-trained models: Comes with models trained on popular music source separation datasets.
• Customizable: Allows users to train their own models using different architectures.
• Lightweight: Optimized for efficient performance on both GPUs and CPUs.
• User-friendly API: Easy integration with existing PyTorch workflows and projects.
• Interactive demos: Provides sample scripts and notebooks to get started quickly.
pip install pytorch-music-source-separation to install the package.Example code snippet:
from pytorch_music_source_separation import separate
mixed_audio = "path/to/mixed_audio.wav"
vocals, instruments = separate(mixed_audio, model="vocal_separation")
vocals.save("vocals_output.wav")
instruments.save("instruments_output.wav")
What operating systems are supported?
PyTorch Music Source Separation is compatible with Windows, macOS, and Linux.
What audio formats are supported?
The tool supports WAV, MP3, and FLAC formats for input and output.
Can I train my own model?
Yes, you can train custom models using your dataset by leveraging the PyTorch framework and the provided training scripts.