Because of the popularity of online multimedia videos, there has been much interest in recent years in acoustic concept detection and classification for the improvement of online video search. In this paper, we present our acoustic concept annotation effort in user-submitted quality videos. We decide on the acoustic concept categories so that their labels can be used as low-level features to distinguish higher-level video events in a multimedia event detection (MED) task. Using thorough acoustic concept annotations, we train acoustic concept models and use these to generate features for a MED task in the form of what we refer to as a segmental GMM-based co-occurrence feature vector.