• DocumentCode
    2192784
  • Title

    Study of speech intelligibility in noisy enclosures using optimal spherical beamforming

  • Author

    Peled, Yotam ; Rafaely, Boaz

  • Author_Institution
    Dept. of Electr. & Comput. Eng., Ben-Gurion Univ. of the Negev, Beer-Sheva, Israel
  • fYear
    2008
  • fDate
    3-5 Dec. 2008
  • Firstpage
    285
  • Lastpage
    289
  • Abstract
    Reverberation and noise have a significant effect on the intelligibility of speech in rooms. Detection of clear speech in highly reverberant and noisy enclosures is an extremely difficult problem. In this work we take the approach of spatial filtering by spherical microphone array to overcome noise and reverberation. In order to investigate speech intelligibility, an acoustic model was developed. The model is based on U50, which was shown to have good correlation with speech intelligibility, while combining the effect of both reverberation and noise. This paper presents the array beamforming design for optimal speech intelligibility performance. The beamformer is designed to enhance the speech signal while minimizing noise power and maintaining robustness at a wide frequency range. Having such a model allows the prediction of the required array order for detecting speech clearly.
  • Keywords
    array signal processing; microphone arrays; spatial filters; speech enhancement; speech intelligibility; optimal spherical beamforming; reverberation; spatial filtering; speech detection; speech intelligibility; speech signal enhancement; spherical microphone array; Acoustic noise; Acoustic signal detection; Array signal processing; Filtering; Frequency; Microphone arrays; Noise robustness; Reverberation; Signal design; Speech enhancement; MVDR; Speech intelligibility; reverberation; room acoustics; spherical microphone array;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Electrical and Electronics Engineers in Israel, 2008. IEEEI 2008. IEEE 25th Convention of
  • Conference_Location
    Eilat
  • Print_ISBN
    978-1-4244-2481-8
  • Electronic_ISBN
    978-1-4244-2482-5
  • Type

    conf

  • DOI
    10.1109/EEEI.2008.4736706
  • Filename
    4736706