Winning Streak Continues: Fifth Award in a Row for the International Audio Laboratories Erlangen

19.03.2015 Link

The International Audio Laboratories Erlangen (AudioLabs) has received its fifth prestigious honour within a short period for its outstanding scientific work related to spatial audio signal processing.

Prof. Dr. Emanuël Habets, leading scientist in the field of spatial audio research at the AudioLabs, has been honoured with the high-profile “2014 Signal Processing Letters Best Paper Award” for his paper, entitled “Late Reverberant Spectral Variance Estimation Based on a Statistical Model” and co-authored by Sharon Gannot (Bar-Ilan University, Israel) and Israel Cohen (Technion – Israel Institute of Technology, Israel).

The paper addresses the challenges caused by reverberant environments, where reflections from walls and other objects mask the sound of direct speech and can therefore degrade the overall speech intelligibility. In particular, these environments may pose significant challenges for people with hearing impairments and/or are using hearing aid systems. Phone calls made via a hands-free system can pose a similar challenge, with reverberation issues increasing the required listening effort and diminishing intelligibility.

In a bid to overcome these long-standing issues, the authors developed an approach to estimate the amount of reverberation using a novel statistical model that depends on two parameters that characterize the reverberant environment: the reverberation time, and the ratio between the direct sound energy and reverberant energy. Based on this estimate, reverberation can be suppressed using an existing speech enhancement method. The proposed approach can be used in speech communication systems to reduce the listening effort and increase the speech intelligibility.

The award-giving ceremony will take place at the 40th IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) 2015 from April 19th to 24th in Brisbane, Australia.

The IEEE Signal Processing Letters Best Paper Award is given by the IEEE Signal Processing Society to authors who have made significant contributions on a notable scientific subject. The selected recipients are characterized by technical innovation, novelty, scientific excellence, and importance for science and research.

More information can be found on the website of the IEEE Signal Processing Society.