Automatic Speech Activity Detection, Source Localization, and Speech Recognition on the Chil Seminar Corpus