Abstract
To meet the security demand of audio information, a generic method of recognizing specific type audio stream based on MFCC and AdaBoost is proposed in this paper, which can detect and locate the specific audio fragment from the audio stream accurately. The generality and differences between subcategories of the audio stream was analyzed to achieve the generic recognition. Multi-type gunshot was considered as the specific type of audio stream. The generic template was obtained by extracting the common features and reducing the different features of the gunshot audio, which could support the accurate identification of multiple sub-classes. The experiments show that the recognition accuracy of the proposed method is 87.6% and the recall rate reaches 91.8%.
Original language | English |
---|---|
Pages (from-to) | 1231-1235 |
Number of pages | 5 |
Journal | Beijing Ligong Daxue Xuebao/Transaction of Beijing Institute of Technology |
Volume | 31 |
Issue number | 10 |
Publication status | Published - Oct 2011 |
Keywords
- AdaBoost
- Audio retrieval
- Mel frequence cepstrum coefficient (MFCC)
- Specific audio stream