A machine learning application for emotion recognition from speech
A machine learning application for emotion recognition from speech.
Language: Python 2.7
Mario Ruggieri
E-mail: [email protected]
Download Berlin DB from the link. Request DaFeX dataset following the link instructions. The code will generate automatically .wav files
Long option | Option | Description |
---|---|---|
--dataset | -d | dataset type |
--dataset_path | -p | dataset path |
--load_data | -l | load dataset data and info and save them into a .p file |
--extract_features | -e | extract features from data and save them into a .p file |
--speaker_indipendence | -s | cross validation is made using different actors for train and test sets |
--plot_eigenspectrum | -i | show eigenspectrum for each training set |
Example:
python emorecognition.py -d 'berlin' -p [berlin db path] -e -l
The first time you run the application, -l and -e options are mandatory because you need to extract data and features. Every time you change the feature extraction method and/or the dataset data you need to specify -e and/or -l to update your .p files.
Please read LICENSE file.