A deep neural net to classify music by mood

Dependencies:

Please install Miniconda3 to create a virtual environment for this project. This will make package conflicts unlikely.

To install Miniconda please go to https://conda.io/miniconda.html

If it warns you about adding Miniconda to your path, make sure you add it!

You can replicate the current Mood Anaconda environment by cloning this repo, then running

conda env create -f environment.yml

Running the live demo

Make sure you have follow the Dependencies instructions above.

In the top level of this repo, simply do

• jupyter notebook.
• Click on the notebooks dir
• Open livedemo.ipynb

And have fun!

SpectrogramCNN

Dependencies

Inside the mood_algorithm/spectrogram-mood-classifier directory you will find a requirements.txt file that lists all the necessary dependencies.

Prediction

See the main() function in /mood_algorithm/spectrogram-mood-classifier/predict.py for example local and cloud predictions. To predict on a bunch of songs using our current CloudML model, simply loop through all your songs, and pass their file names into the cloud_predict() function.