According to result of the 6th?Detection and Classification of Acoustic Scenes and Events 2020 (DCASE 2020) Challenge organized by the Technical Committee of Audio and Acoustic Signal Processing (AASP) of Institute of Electrical and Electronics Engineers (IEEE), Maxvision Technology, which participated in the event for the first time, won the domestic second place and the international fourth place in the Task 1a ¡°Cross-device Acoustic Scene Classification¡± task as a team.?

DCASE Challenge is a specialized event initiated in 2013 by Queen Mary University of London (QMUL) and focusing on acoustic scene classification, acoustic event detection and recognition and so on. It is the most influential international event in the field of audio and acoustic signal processing.?

DCASE 2020 has the following 6 tasks: acoustic scene classification, anomalous sound detection, acoustic event detection, acoustic event separation, urban sound tagging and automated audio captioning. 138 teams with 473 effective systems from the world participated in it. Task 1a ¡°cross-device acoustic scene classification¡± is the task with the fiercest competition and the most participants.?

Compared with DCASE 2019, DCASE 2020 had a Task 1a data set contained audios recorded by different devices, meeting the practical application demands more and making the challenge much more difficult. The task attracted 27 top acoustic teams from the academic community and the industry community in the world, receiving 71 effective results submitted. In the competition with the world¡¯s well-known enterprises and research institutions such as Tencent multimedia lab, Intel and ETRI, and domestic and foreign top universities and colleges such as Tsinghua University, The Chinese University of Hong Kong, University of Electronic Science and Technology of China, and Seoul National University, we won the domestic second place and the international fourth place (only 0.2% lower than the third place) as a team.

?

The system submitted by us, which is mainly used for the effective sound analysis, enhancement processing and intelligent identification by our intelligent service robot system and other man-machine interaction products in specific scenes, was developed by our AI research team. The AI research team has long been engaged in the technical researches of AI, deep learning, acoustic scene recognition, audio signal processing, image processing and so on in order to apply the cutting-edge AI algorithms more quickly. Multiple research achievements of it have been widely used in the smart port field. We will keep developing AI applications in sound signal processing and sound event detection.