A project indulged with generating ear soothing sounds using artificial neural network. This project generated sounds on based of user request with various different instrument .
The model is based on image generation using W-GAN but instead of using real images we use a binary representation of beats, and stop duration on an image and use our W-gan to generate that bit image, and then we transfer that image back to original beats, Notes and chords
pip install -r requirements.txt
- once setup is done you need to download models for each instrument and place it under weights folder DRUM ,PIANO
- Currently this script is able to render 2 different instrument i.e. Piano and Drum {0 for drum,1 for piano}
- Demo command to get user started
python melody_generator.py --instrument 0 --sequence 3
- After running the above command you will get two output one bit image and second midi file play it and enjoy <:)
Drum model weights updated in drive //TODO train model on more instruments
Piano model weights updated in drive //TODO train model on more instruments