The official code repo for "Zero-shot Audio Source Separation through Query-based Learning from Weakly-labeled Data", in AAAI 2022

A local GUI music source separation tool built on Tkinter and demucs serving as a free and open source Stem Player

A Python toolkit for sound source separation.

Code accompanying the paper "Semi-supervised adversarial audio source separation applied to singing voice extraction"

logWMSE, an audio quality metric & loss function with support for digital silence target. Useful for training and evaluating audio source separation systems.

Variations of L1 SNR Loss function for training audio source separation machine learning models

Karaokey is a vocal remover that automatically separates the vocals and instruments. A deep learning model based on LSTMs has been trained to tackle the source separation. The model learns the particularities of music signals through its temporal structure.

REPeating Pattern Extraction Technique (REPET) in Matlab for audio source separation: original REPET, REPET extended, adaptive REPET, REPET-SIM, REPET-SIM online

REPeating Pattern Extraction Technique (REPET) in Python for audio source separation: original REPET, REPET extended, adaptive REPET, REPET-SIM, online REPET-SIM

Piano Concerto Source Separation Pipeline

Spectral Mapping of Singing Voices: U-Net-Assisted Vocal Segmentation

all-in-one ai-midi-pipeline

Audio Source Separation using the Non Negative Matrix Multiplication

Universal frequency-band splitter for music source separation: one interface for BSRoformer, MelBandRoformer, or any Band Split. RoPE, PoPE, HyperACE. Documentation, citations, and explanations. PyTorch; CUDA-accelerated.

Matlab GUI for uREPET, a simple user interface system for recovering patterns repeating in time and frequency in mixtures of sounds.

Testing The Efficiency of WaveUNet as an Audio Source Separation model

Matlab GUIs to demo the original REPET and REPET-SIM.

Neural network model for audio source separation task

AI pipeline detecting infrasonic elephant rumbles (8–50 Hz) in noisy field recordings. Three-path DSP detector (pYIN voicing + harmonic contrast + F0 stability) + U-Net denoiser + Wiener-mask caller separation. Built at HackSMU VII for ElephantVoices.

Prototype AI-driven web application that analyzes and transcribes musical compositions by extracting notes and chords from audio files. The system employs advanced machine learning and audio processing techniques.