Protocol and baseline for experiments on Bogazici University Turkish emotional speech corpus

This study aims at presenting an emotional corpus collected at Boğaziçi University / Electrical and Electronics Department, on which no previous signal processing and machine learning study was done for classification purposes. It also aims at providing the protocol for further experiments on this corpus. The emotional corpus consists of 484 speech utterances from 11 amateur actors acting 11 emotionally undefined sentences with 4 emotions using Stanislavsky effect. In-line with the state-of-the-art method in the field, functionals were passed over the Low Level Descriptors of the signal to obtain fixed length feature vectors. For this purpose, the openSMILE feature extractor was used with the baseline feature set from the INTERSPEECH 2013 Computational Paralinguistics Challenge. The training, validation and testing partitions are defined on the corpus. The baseline results obtained using Support Vector Machines and Random Forests are presented.