- project
1.1 data
1.1.1 fma_metadata - consists of the metadata needed to load audio files (tids)
1.1.2 fma_small - the small dataset downloaded from the repository given in data.txt
1.1.3 npy_files - contains processed files for medium and small datasets (download from link given in data.txt)
1.1.4 weights - our trained weights are stored in crnn_20_latest.h5
1.2 docs - contains graphs, report and some logs
1.3 fma_env - virtual env for the project
1.4 presentation - contains the project presentation
1.5 sources - various research paper and other sources used in the project
1.6 src
1.6.1 helper - contains some helper files
1.6.2 processing - contains .py programs to help with pre-processing, cleaning metadata, conversion to npy
1.6.3 trials - contains experimental configs of the network
1.6.4cnn_model_using_npy_medium_set.ipynb
is the core file that loads the npy files and trains the network
1.6.5visualize.ipynb
is used in producing visuals for the network
- The config variales at the beginning of the files must be changed to use either small or medium datasets for training, plotting etc