.
├── README.md
├── base-model.ipynb # train model
├── emotion-gender.ipynb # evaluation with gender
├── emotion-without_gender.ipynb # evaluation without gender
├── gender100.pth # pretrain model with gender 
├── 100.pth # pretrain model without gender 
└── testdata

• python3
• librosa
• numpy
• torch
• sklearn
• pandas

For dataset, you could download RAVDESS or use records in /testdata to test code.

Modify the path in /base-model.ipynb , change the path in [] into the path where you download the dataset.

path ='[../input/ravdess/RAVDESS]'
...
file_path.append('[../input/ravdess/RAVDESS/]' + file)

You could change the training epoch in /base-model.ipynb

for epoch in range(1,60):

For dataset, you could download RAVDESS or use records in testdata to test code. Modify the path in /emotion-without_gender.ipynb or /emotion-gender.ipynb , change the path in [] into the path where you download the dataset.

path ='[../input/ravdess/RAVDESS]'
...
file_path.append('[../input/ravdess/RAVDESS/]' + file)

You could change the training epoch in /emotion-without_gender.ipynb or /emotion-gender.ipynb

for epoch in range(1,100):

In this files, output of confusiom matrix, classification report and emotion count report are all saved. If you want to retest, just run the files. Or preload our pretrained model, for no-gender the pretrained model is named as 100.pth and with gender is gender100.pth