Talk on Distributed Signal Processing by Richard Heusdens

As part of the annual workshop on the Villum project, Richard Heusdens will give a one-hour  introduction to distributed signal processing at 9:20 on April 23.

Title: Distributed signal processing

Abstract: In this talk we will focus on distributed signal processing algorithms. We will discuss methods based on inference in graphical models, like the mini-sum and related algorithms, and methods based on convex optimization, like the alternating direction of multipliers method (ADMM) . With respect to inference-based algorithm, we will focus on the (generalized) linear-coordinate descent algorithm, an iterative optimization algorithm with a convergence rate comparable to that of the min-sum algorithm, but with significantly less parameters to transmit per iteration. With respect to convex optimization based algorithms, we will focus on ADMM, a simple but powerful algorithm that is well suited for distributed convex optimization, and variations of it, like the recently introduced BiADMM algorithm, which can be used for both synchronous and asynchronous updating schemes.

picture-21Bio: Richard Heusdens is an Assoc. Prof. at the Dept. of Mediamatics, Delft University of Technology. He received his HTS (institute of technology) diploma from the HTS Alkmaar in 1988. He received the M.Sc. and Ph.D. degrees from Delft University of Technology, Delft, The Netherlands, in 1992 and 1997, respectively. Since 2002, he has been an Associate Professor in the Department of Mediamatics, Delft University of Technology. In the spring of 1992, he joined the digital signal processing group at the Philips Research Laboratories, Eindhoven, The Netherlands. He has worked on various topics in the field of signal processing, such as image/video compression and VLSI architectures for image processing algorithms. In 1997, he joined the Circuits and Systems Group of Delft University of Technology, where he was a Postdoctoral Researcher. In 2000, he moved to the Information and Communication Theory (ICT) Group, where he became an Assistant Professor responsible for the audio and speech processing activities within the ICT group. He is involved in research projects that cover subjects such as audio and speech coding, speech enhancement, and digital watermarking of audio. Research projects he is involved in cover subjects such as audio and speech coding (Sicas, ARDOR, ASC), speech enhancement (SpEnt), and digital watermarking of audio (DIWAMETRICS)

New M.Sc. program in Sound and Music Computing at AAU

smcbannerIn September 2014, AAU will start a new M.Sc. program in Sound and Music Computing. The program features a unique combination of traditional engineering disciplines like signal processing and machine learning with newer ones like interaction design and prototyping and fabrication techniques and strong knowledge about various aspects of sound and music, including perception and cognition, acoustics, psycho-acoustics. The new program will run both in Aalborg and in Copenhagen and is open for both national and international applicants from a wide range of B.Sc. programs. The Audio Analysis Lab has been heavily involved in the creation of the program with our colleagues in Copenhagen and will also be involved in teaching courses and supervising students in the program. You can read more at http://www.studieguide.aau.dk/uddannelser/kandidat/93920/.

Hendrik Purwins has joined the lab

We are happy to announce that a recent hiring of AD:MT, Assistant Professor Hendrik Purwins, has joined the Audio Analysis Lab! We welcome Hendrik to the fold and look forward to years of fruitful collaborations.

citationsAfter concluding his studies in mathematics at Bonn and Münster universities, Hendrik Purwins was awarded a Ph.D. on machine learning and musical signal processing with a scholarship of the Studienstiftung des deutschen Volkes (given only to the best 0.5 % students at German universities) after studying at Berlin Institute of Technology (BIT), CCRMA, Stanford, and the Psychology Dept. of McGill University. Since, he has been guest professor, senior researcher, and lecturer at Music Technology Group at Pompeu Fabra University, Barcelona, and guest researcher at IRCAM, Paris, and Neurotechnology Group / Berlin Brain Computer Interface, BIT. In 2013, he became Assistant Professor at Aalborg University Copenhagen. He has (co-)authored more than 70 scientific papers, has received 10 personal research grants and prizes, and has lead research teams in 3 European projects. Starting with playing the violin at age of 7, he has played several string quartet and orchestra concerts during his mandatory military service with the professional German head quarters chamber orchestra. His research interests include signal processing, machine learning, and experimental methods (psychological and EEG experiments) for models of sound and music and their application to resynthesis, visualization, user adaptation, and non-Western music.

Book on speech enhancement out now!

downloadWe are proud to announce that our recent results on speech enhancement using subspace-based filtering have now been published in a book. The book, entitled “Speech Enhancement — A Signal Subspace Perspective” is published with Academic Press, and is authored by Prof. Jacob Benesty, Postdoc Jesper Rindom Jensen, Prof. Mads Græsbøll Christensen and Prof. Jingdong Chen.

In the book, we are the first to treat subspace approaches in a unified way for time and frequency domain, single-channel, multichannel, as well as binaural, speech enhancement. Moreover, the book bridges the gap between optimal filtering methods and subspace approaches, which has previously been considered as separate fields within speech enhancement.

Postdoc Grant for Jesper Rindom Jensen!

Jesper_Rindom_Jensen_bw-300x231We are extremely pleased and proud to announce that Audio Analysis Lab founding member Jesper Rindom Jensen has received one of the very competitive postdoc grants from The Danish Council for Independent Research! Only 24 out of a total of 170 grant proposals received funding. The project is entitled Localization and Tracking of Speech – A Joint Audio-Visual Approach and receives approximately 2.6 mio DKK in funding! Read more here.

Jacob Benesty appointed Adjunct Professor

The Audio Analysis Lab is extremely pleased to announce that the Academic Council at the Faculty of Engineering and Science has decided to confer the title of adjunct professor to Jacob Benesty for a period of 5 years from January 1 2014. He will be part of our lab and will work closely with us to expand the research activities in noise reduction and microphone array signal processing.

benestyJacob Benesty received his M.Sc. from Pierre & Marie Curie University, Paris France in 1987 and his Ph.D. (summa cum laude) from Orsay University, Paris France in 1991. He is currently Professor at INRS-EMT, University of Quebec, QC, Montreal, Canada. Before joining INRS-EMT, he was Member of Technical Staff in the Acoustics Research Department at Bell Laboratories from 1995 to 2003. He has published no less than 18 books, 103 journal papers, 33 book chapters, 153 conference papers and is inventor of 14 granted patents, all on the topics of adaptive filtering, microphone array signal processing, and speech enhancement. Moreover, he has received several best paper awards, including two prestigious IEEE Signal Processing Society Best Paper Awards for his contributions. His high impact is evident from the very high number of citations his work has received. According to Google Scholar, he has an h-index of 44, an i10-index of 133 and a total of 8005 citations.

Going to WASPAA

Two of the lab members (Mads Græsbøll Christensen and Jesper Rindom Jensen) are participating in this years IEEE Workshop on Applications of Signal Processing to Audio and Acoustics (WASPAA). WASPAA is held every 2nd year in the Mohonk Mountain House, New Paltz, NY, and is visited by the leading experts within audio signal processing.

During the workshop, the lab will present the work entitled:

  • An LCMV Filter for Single-Channel Noise Cancellation and Reduction in the Time Domain.

The work was carried out in collaboration with Prof. Jacob Benesty and Prof. Jingdong Chen.

See you in New Paltz!

EUSIPCO Papers Accepted

The Audio Analysis Lab had two out of two papers accepted for presentation at EUSIPCO 2013:

  • Fast Joint DOA and Pitch Estimation Using a Broadband MVDR Beamformer
  • On the Influence of Inharmonicities in Model-Based Speech Enhancement

The papers were written by Sam Karimian-Azari and Sidsel Nørholm (with  co-authors) who are funded by the Villum Foundation. Congratulations to Sam and Sidsel on work well done!

EUSIPCO 2013, i.e., the 21st European Signal Processing Conference, is held in September in  Marrakech in Morocco.

Villum Foundation Project Workshop 2013

On May 15, the first workshop on the project Spatio-Temporal Filtering Methods for Enhancement and Separation of Speech Signals was held. The program consisted of the following presentations:

  • A. Jakobsson “Estimation of Multiple Pitches using Block Sparsity”
  • J. K. Nielsen “Model Comparison with the g-Prior”
  • S. I. Adalbjörnsson “Robust Fundamental Frequency Estimation in the Presence of Inharmonicities”
  • S. M. Nørholm “Inharmonicities in Model-Based Speech Enhancement”
  • J. R. Jensen “Joint DOA and Fundamental Frequency Estimation Based on Relaxed Iterative Adaptive Approach and Optimal Filtering”
  • S. Karimian-Azari “Fast Joint DOA and Pitch Estimation Using a Broadband MVDR Beamformer”
  • S. M. Nørholm “Noise Reduction using Joint Diagonalization”
  • J. R. Jensen “Multichannel Signal Enhancement using Non-Causal, Time- Domain Filters“
  • M. G. Christensen “An Exact Subspace Method for Fundamental Frequency Estimation”

The participants were Jacob Benesty, Andreas Jakobsson, Mads Græsbøll Christensen, Jesper Rindom Jensen, Jesper Kjær Nielsen, Stefan Ingi Adalbjörnsson, Sam Karimian-Azari, Sidsel Marie Nørholm, and Søren Vang Andersen.