Moshirynia - REVERB Workshop 2014 A SPEECH DEREVERBERATION...

Info icon This preview shows pages 1–2. Sign up to view the full content.

View Full Document Right Arrow Icon
A SPEECH DEREVERBERATION METHOD USING ADAPTIVE SPARSE DICTIONARY LEARNING M. Moshirynia, F. Razzazi, A. Haghbin* Department of Electrical and Computer Engineering IAU, Science and Research Branch Tehran, Iran [email protected], [email protected], [email protected]* ABSTRACT We present a monaural blind de-reverberation method based on sparse coding of de-convolved version of reverberated speech signal in a dictionary which is learned by joint dictionary learning method, consisting of the concatenation of a clean speech and a non-negative matrix factor de- convolution result of the reverberated copy. The environment specific dictionary is originally learned off-line on a training corpus for different locations, while adaptive dictionary learning continues on-line for any other surroundings. Our approach uses both non-negative blind de- convolution and sparse coding, and achieves some improvements on objective voice quality testing’s like perceptual evaluation of speech quality. Index Terms —speech dereverberation, sparse coding, dictionary learning, Non-negative Matrix Factor Deconvolution 1 INTRODUCTION Dereverberation is a highly appropriate and difficult task. Its importance is due to the variety of practical applications while the difficulty arises from the dynamic and long time effect that reverberant environments influence on speech signals. The aim of adaptive dictionary learning is to remove long and non-stationary facts which significantly reduce both speech quality and intelligibility. Dereverberating speech which is degraded by convolutional noise of reverberant environments (e.g. classrooms, halls, inside car, etc) is both a highly quality improving and difficult task. Its importance is due to its contribution in quality of service in various practical signal processing applications including mobile communications, speech recording and speech recognition. The difficulties are from the dynamic and long time degradation that reverberation affects on speech signals. Technically speaking, reverberation is the effect of the acoustic channel from the speech source up to the hearing system. The effect of reverberation begins with the production of sound at a location inside a room. The acoustic wave expands radial, reaching walls and other surfaces where energy is both absorbed and reflected. This effect can mainly be modeled by a linear time invariant system: ± ² [³]= ± ´ [³]∗ ℎ[³] (1) where, * denotes discrete time linear convolution, s c [ n ] is source or clean speech signal, h [ n ] is the impulse response of a linear system or RIR (Room Impulse Response), s r [ n ] is the reverberated signal and n is the time index. The parameters of the filter h [ n ] change with changes in the environmental parameters such as size of the room, room configuration, position of objects etc. It is typically assumed that the room-response spectral variations rate is slow comparing to the spectral variation rate of speech. As a result, for a short duration (e.g. two or three seconds), we can assume that h [ n
Image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
Image of page 2
This is the end of the preview. Sign up to access the rest of the document.

{[ snackBarMessage ]}

What students are saying

  • Left Quote Icon

    As a current student on this bumpy collegiate pathway, I stumbled upon Course Hero, where I can find study resources for nearly all my courses, get online help from tutors 24/7, and even share my old projects, papers, and lecture notes with other students.

    Student Picture

    Kiran Temple University Fox School of Business ‘17, Course Hero Intern

  • Left Quote Icon

    I cannot even describe how much Course Hero helped me this summer. It’s truly become something I can always rely on and help me. In the end, I was not only able to survive summer classes, but I was able to thrive thanks to Course Hero.

    Student Picture

    Dana University of Pennsylvania ‘17, Course Hero Intern

  • Left Quote Icon

    The ability to access any university’s resources through Course Hero proved invaluable in my case. I was behind on Tulane coursework and actually used UCLA’s materials to help me move forward and get everything together on time.

    Student Picture

    Jill Tulane University ‘16, Course Hero Intern