Abstract | ||
---|---|---|
Sound event detection (SED) aims to identify the types and the temporal boundaries of the target sound events. Successful detection of various sound events can reveal rich information in many home and surveillance applications. However, this task is challenging due to the complicated realworld background noise and the overlapping of sound events. In this paper, we proposed a unified approach that takes the advantage of both deep learning and audio enhancement. A convolutional recurrent neural network (CRNN) is combined with a deep neural network (DNN) to improve the performance of the SED classifiers, and an optimally modified log-spectral amplitude estimator (OMLSA) based audio enhancement method is employed to improve the robustness of the SED system on the noisy environment. Experiments on several datasets show a more than 3% increase in accuracy in the normal environment and a more than 9% increase in accuracy in the noisy environment compared to the state-of-the-art approaches. |
Year | DOI | Venue |
---|---|---|
2019 | 10.1109/ISSPIT47144.2019.9001843 | 2019 IEEE International Symposium on Signal Processing and Information Technology (ISSPIT) |
Keywords | DocType | ISSN |
sound event detection,OMLSA,CRNN,DNN,audio enhancement | Conference | 2162-7843 |
ISBN | Citations | PageRank |
978-1-7281-5342-1 | 0 | 0.34 |
References | Authors | |
9 | 4 |
Name | Order | Citations | PageRank |
---|---|---|---|
Tongtang Wan | 1 | 0 | 0.34 |
Yi Zhou | 2 | 15 | 9.83 |
Yongbao Ma | 3 | 0 | 0.68 |
Hongqing Liu | 4 | 45 | 28.77 |