sound gathering and preprocessing, detector training and testing. The system can recognize
thirteen types of sound events, such as baby crying, a person screaming or asking for help
and so on. It has extremely low false positive rate and can be used for automated continuous
monitoring.