| Paper: | AE-P4.11 | ||
| Session: | Applications to Music | ||
| Time: | Thursday, May 18, 16:30 - 18:30 | ||
| Presentation: | Poster | ||
| Topic: | Audio and Electroacoustics: Applications to Music | ||
| Title: | Frequency Component Restoration for Music Sounds Using a Markov Random Field and Maximum Entropy Learning | ||
| Authors: | Tomonori Izumitani, Kunio Kashino, NTT Corporation, Japan | ||
| Abstract: | Restoring frequency components hidden by other interfering sounds is a difficult problem but has become important in various music information processing systems. We propose a method that estimates and restores frequency component structures from music signals with noise. It is based on a Markov random field (MRF) model constructed with a supervised learning technique using the maximum entropy model. It is shown that the method achieves F-measures greater than 0.44 even in periods where signals are completely replaced by noises. Moreover, we evaluated the method in terms of feature distortion recovery in audio fingerprint matching tasks. The results show that the proposed method clearly reduces the effect of noise on the similarity values. | ||
©2018 Conference Management Services, Inc. -||- email: webmaster@icassp2006.org -||- Last updated Friday, August 17, 2012