• DocumentCode
    2575340
  • Title

    A MDCT-based click noise reduction method for MPEG-4 AAC codec

  • Author

    Chen, Hao ; Bao, Changchun ; Deng, Feng ; Zhang, Dawei ; Jia, Maoshen

  • Author_Institution
    Speech & Audio Signal Process. Lab., Beijing Univ. of Technol., Beijing, China
  • fYear
    2011
  • fDate
    9-11 Nov. 2011
  • Firstpage
    1
  • Lastpage
    5
  • Abstract
    A MDCT-based click noise reduction approach used for MPEG-4 AAC codec is proposed in this paper. This noise reduction technique can be embedded in MPEG-4 AAC codec as a function module of audio restoration. The proposed method contains two parts: click noise detection and click noise removal. The click noise detection is further divided in two steps. Firstly, the window type for MDCT is decided by perceptual entropy (PE). Secondly, the spectral entropy (SE) of clean audio signal and noisy signal can be calculated and the position of clicks noise can be determined using minimum error Bayesian decision. A MAD-based soft-thresholding criterion is adopted for click noise removal. In addition, the spectral smoothing is employed in the adjacent frames. The test results indicate that the proposed algorithm can improve the audio quality and provide a comparable performance with conventional click noise method.
  • Keywords
    Bayes methods; audio coding; codecs; discrete cosine transforms; signal denoising; MAD-based soft-thresholding criterion; MDCT-based click noise reduction method; MPEG-4 AAC codec; audio quality; audio restoration; click noise detection; click noise removal; function module; minimum error Bayesian decision; spectral smoothing; window type; Codecs; Entropy; Noise; Noise measurement; Noise reduction; Transform coding; Transient analysis; MDCT; click noise reduction; perceptual entropy; soft thresholding; spectral entropy;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Wireless Communications and Signal Processing (WCSP), 2011 International Conference on
  • Conference_Location
    Nanjing
  • Print_ISBN
    978-1-4577-1009-4
  • Electronic_ISBN
    978-1-4577-1008-7
  • Type

    conf

  • DOI
    10.1109/WCSP.2011.6096795
  • Filename
    6096795