Skip to content

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.

License

Notifications You must be signed in to change notification settings

NotVinay/karaokey

Repository files navigation

Karaokey

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.

The model is trained using PyTorch and deployed using Flask.

Dataset used to train the model

The models are trained using the MUSDB18 dataset, here is the link to MUSDB18 dataset. More information about the dataset can also be found at sigsep.io.

Installation

Use the package manager pip to install all dependencies needed for Karaokey.

Installing libsndfile and ffmpeg

Libsndfile and ffmpeg is needed for audio processing.

For linux run this command

sudo apt install -y libsndfile1 ffmpeg 

For windows To install libsndfile and ffmpeg we will have to use conda environment alternatively you can also follow the instructions on libsndfile.

conda install -c conda-forge ffmpeg libsndfile

For latest dependencies run

pip install -r requirements.txt

For stable dependencies run

The versions mentioned in this files are tried and tested. The application was tested on this versions.

pip install -r requirements-versions.txt

For latest dependencies for gpu run

If you want to train the models on GPU please install these dependencies. Cuda based GPU dependencies are mentioned in this file, however if you want to install any other ones please have a look at PyTorch's installation guidelines.

pip install -r requirements-gpu.txt

Note: If you get any error while running the scripts of application, please try and install the versions used "requirements-versions.txt" the versions mention in this file is tried and tested.

Running the Flask Application

If you want to run Karaokey application using the trained model. Please follow following steps. Make sure you have all dependencies installed. Setting the flask environment variables.

env FLASK_APP=./application/__init__.py 

Run flask app

flask run

About

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.

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published