Audio context recognition using audio event histograms

This paper presents a method for audio context recognition, meaning classification between everyday environments. The method is based on representing each audio context using a histogram of audio events which are detected using a super-vised classifier. In the training stage, each context is modeled with a histogram estimated from annotated training data. In the testing stage, individual sound events are detected in the unknown recording and a histogram of the sound event oc-currences is built. Context recognition is performed by com-puting the cosine distance between this histogram and event histograms of each context from the training database. Term frequency–inverse document frequency weighting is studied for controlling the importance of different events in the his-togram distance calculation. An average classification accu-racy of 89 % is obtained in the recognition between ten every-day contexts. Combining the event based context recognition system with more conventional audio based recognition in-creases the recognition rate to 92%. 1