Best features for emotional speech classification in the presence of babble noise

Hitherto, different efforts have been held for the recognition of emotional state of speakers. Most of these works are performed in clean environments. But, in the real world, there are different noise parameters such as cross-talk, car noise, awgn (especially in the transmission of sounds) and etc., which decrease the performance of classifiers. In this paper we look for features which have the best performance in the presence of babble noise. We carry out our evaluation on three emotional speech datasets.