Acoustic Event Detection and Localization

The Acoustic Event Detection Support Vector Machine-based system (Temko et al., CLEAR 2007), written in C++ programming language, is part of the smartAudio++ software package developed at TALP Research Center, Speech Processing Group, Universitat Politèctnica de Catalunya (UPC, Barcelona, Spain) which includes other audio technology components such as speech activity detection and speaker identification for the purpose of real-time online activity detection and observation in the smart-room environment. That AED system implemented in the smart-room has been used in the demos about technology services developed in CHIL project (Computer in the human interaction loop). Also, a specific GUI-based demo has been built which shows the detected isolated events and the positions of their sources in the room. The positions are obtained from the acoustic source localization (ASL) system developed also in our lab (Segura et al., CLEAR 2007). A video recording has been captured from the GUI and is publicly available. It contains the output of the GUI during a session lasting about 2 minutes, where four people in the room speak, are silent, or make one of the 12 meeting-room sounds defined in CHIL and a few others.

Category:

Science & Technology

Tags:

• Detection
• Acoustic
• Event
• AED
• Support
• Vector
• Machines
• UPC

[devkarmani]

hey great video ....I am also working on this area ...could you please identify yourself to me ...may be you can help me in many ways...thanks..bye the good work..I am MSc student.

[andtem2000]

google Andrey Temko

[m4rkus123]

I see that only one person make noise. This is limit of this program to one source of sound? Or it could work with many sound source? What kind of source localization and what kind of event detect use in your application?