Goals Accomplished This week:
- Data Normalization was completed.
- Feature extraction through two configuration files IS09 and Avec13 was done.
- The libsvm parameters was obtained.
- The test of SVM algorithm according to the principle of leaving one speaker out was finished.
- Decision tree method was tried by Python sklearn.
Goals to be Accomplished Next Week :
- Determine whether SVM is not appropriate to our project.
- Try other ways to do the classification.
Technical Difficulties:
- Accuracy of result processed by SVM was extremely low.
- The more features extracted the lower the accuracy to be with SVM.
- The proper classification algorithm can not be figured.
Weekly Report