Music generation using Deep Learning
Software solution that generates musical melodies according to user preferences and requirements using Deep Learning.
Music is called a universal language because music communicates feelings and emotions in forms of rhythm, melody, harmony, and timbre. Music composition is done by humans with their creativity, and these composers own their music. Composing of music is not an easy procedure. It requires a deep understanding, knowledge, and artistic sense.
Musc is using a deep learning approach to tackle this problem. With that, a desired music can be generated according to the preferences and requirements of the user.
- Melody Generation: Choose model, customize duration, tempo, temperature and save generated melodies.
- Model Generation: Upload custom datasets, generate from scratch or finetune models, and manage models.
- View History: View timestamped melodies, playback, save, and delete from history.
Follow these steps to set up and run Musc on your local machine:
-
Clone the Repository:
git clone https://github.com/DinukaGayashan/Musc.git cd musc -
Add Pre-trained models: (Optional) Download and place pre-trained models at
models/trained_models. Available here. -
Install Python: Make sure Python 3.10 is installed. Download it from python.org.
-
Install Requirements:
pip install -r requirements.txt
-
Run the Application:
streamlit run main.py
Or else to run with Docker by replacing steps 3 to 5:
-
Install Docker: Make sure Docker is installed. Download it from docker.com.
-
Build Docker Image:
docker build -t musc . -
Run with Docker:
docker run -p 8501:8501 musc
Samples available at SoundColud.
This project is highly inspired by Pop Music Transformer: Beat-based Modeling and Generation of Expressive Pop Piano Compositions
Feel the magic of music with Musc!