This is a small machine learning related project dealing with the Cold sub-challenge proposed in the INTERSPEECH 2017 ComParE Challenge.
You could download the data from INTERSPEECH website.
First run GenerateGMM.py and followed by SubdicFV_dict.py to generate encoded feature. Finally run the StrengthModeling.py to generate the result of different combination of encoded feature.
To see more detail of approach please refer to the blog in my personal website.
Thanks to all teammates, Hao-Chun Yang, Jen-Lin Li, Selly Lu, Bo-Hao Su and Eileen Shieh, work so hard within less than a month to finish this small project.