The Audio Analysis Lab’s work on diagnosis of Parkinson’s voice from speech signals was featured in an article (in Danish) in the popular scientific magazine Aktuel Naturvidenskab. The article can be found here. The article describes the disease, its treatments, and how the voice is affected by the disease , and how it is possible that the disease can also be detected from the voice. The article specifically mentions our work on vocal tract modeling, fundamental frequency estimation, and the Ph.D. project of Audio Analysis Lab member Amir Poorjam.

Audio Analysis Lab member Mads Græsbøll Christensen was guest in the podcast Athene produced by Folkeuniversitet Aalborg. In the podcast, Mads talks about “Five things you didn’t know you could do with Sound” based on the Audio Analysis Lab’s research. Among the topics covered is how voice analysis can be used for diagnosis of illnesses, how superhuman hearing could be possible with networks of audio devices, how you can hear your acoustic surroundings and how that can be used in robot navigation, how you can define sound zones in a room where you can hear different things, and how ??. You can listen to the podcast (in Danish) here.

On August 16, the Audio Analysis Lab celebrated its 10th anniversary at the annual Audio Analysis Workshop which was funded by the GN Foundation. The lab was founded in 2012 based on a Villum Foundation Young Investigator Programme (YIP) grant received by lab founder and head Mads Græsbøll Christensen and that same year the first workshop was held. The anniversary was celebrated with a talk looking back at the past 10. This year’s workshop featured 17 invited presentations covering a wide range of research topics within audio and acoustic signal processing. The talks covered topics such as optimal filtering, source localization, sound zones, active noise cancellation, impulse response modeling, and speech enhancement. There were 41 participants from KU Leuven, Lund University, Aalborg University and from a number of prominent companies.

As always, the Audio Analysis Lab is well-represented with papers and presentations at ICASSP 2022 (https://2022.ieeeicassp.org/) which will take place virtually May 7-13, 2022 and physically May 22-27, 2022.

• AUD-37.4: GENERATION OF PERSONAL SOUND FIELDS IN REVERBERANT ENVIRONMENTS USING INTERFRAME CORRELATION
• AUD-25.5: COMPUTATIONALLY EFFICIENT FIXED-FILTER ANC FOR SPEECH BASED ON LONG-TERM PREDICTION FOR HEADPHONE APPLICATIONS
• MLSP-42.2: PRIVACY-PRESERVING DISTRIBUTED EXPECTATION MAXIMIZATION FOR GAUSSIAN MIXTURE MODEL USING SUBSPACE PERTURBATION
• AUD-19.4: SPARSE MODELING OF THE EARLY PART OF NOISY ROOM IMPULSE RESPONSES WITH SPARSE BAYESIAN LEARNING
• AUD-11.4: A BAYESIAN PERMUTATION TRAINING DEEP REPRESETNATION LEARNING METHOD FOR SPEECH ENHANCEMENT WITH VARIATIONAL AUTOENCODER
• SS-4.5: ROBUST PRESSURE MATCHING WITH ATF PERTURBATION CONSTRAINTS FOR SOUND FIELD CONTROL

TIME AND PLACE:

The public defense will take place 12:00 – 16:00 o’clock in room 3.107 at CREATE, Aalborg University, Rendsburggade 14, 9000 Aalborg. For more information contact Kristina Wagner Røjen, phone: +45 9940 9926, e-mail: kwro@create.aau.dk.

Sign up for reception: 

Link to doodle
 

ASSESSMENT COMMITTEE:

• Professor Stefania Serafin (Chairwoman), Aalborg University (DK)
• Professor Gaël Richard, TELECOM Paris (FR)
• Professor Patrick Naylor, Imperial College (UK)

PROGRAM FOR THE DEFENSE:

12:00 – 12:15  Welcome by the moderator Dean Henrik Pedersen

12:15 – 12:45  Presentation by Professor Mads Græsbøll Christensen

12:45 – 13:00  Break

13:00 – 16:00  Questions from the assessment committee / ex auditorio

16:00 – 18:00  Reception (sign up) 

INVITATION

ABSTRACT:

This thesis is concerned with model-based analysis and processing of speech and audio signals, to which a number of scientific contributions are made in the form of new mathematical models and new methods for the processing of such signals. The thesis demonstrates how a number of models can be used for modeling speech and audio signals in different ways and for different purposes. It is shown how the problem of estimating the parameters of these models can be solved in a number of principled ways using methods such as maximum likelihood, subspace methods, and sparse approximations, whereby both very accurate and robust estimators that explicitly take the properties of speech and audio signals and the presence of noise into account are obtained. Among the parameter estimation problems considered are those of fundamental frequency estimation, linear prediction, source localization, and order selection, problems that have many important applications in speech and audio processing, including the analysis, coding, and enhancement of such signals. It is then shown how such models can be integrated in filtering methods to solve both signal enhancement and parameter estimation problems, such as noise statistics estimation and fundamental frequency estimation, and it is shown how these principles can be extended to multiple channels to solve the problems of beamforming and source localization. The results of the thesis as a whole demonstrate the benefits of the model-based approach compared to the typically non-parametric methods used in speech and audio processing, not only in terms of obtaining new and better methods but also advancing our understanding of both speech and audio signals and the associated estimation problems.

The past years have been productive for the Audio Analysis Lab with several PhD students defending their Ph.D. theses and graduating:

• Taewong Lee defended his thesis entitled ” The Creation of Perceptually Optimized Sound Zones Using Variable Span Trade-Off Filters” on January 11, 2022.
• Alfredo Esquivel defended his thesis entitled ”
  Pre-processing of Speech Signals for Robust Parameter Estimation ” on November 11, 2021.
• Amir Poorjam defended his thesis entitled “Automatic Quality Control in Speech Data Collection with Application to Remote Pathological Voice Analysis ” on August 16, 2021.
• Qiongxiu Li defended her thesis entitled “Privacy-Preserving distributed processing over networks” on August 30, 2021.

CREATE Ph.D. student Qiongxiu Li, and Audio Analysis Lab member, won the 3 minute thesis contest at EUSIPCO 2020 last week with her presentation entitled Privacy-Preserving Distributed Signal Processing. Below you can see her award-winning presentation. Qiongxiu is part of the interdisciplinary research project SECURE at Aalborg University, which you can read more about here. She is supervised by Prof. Mads Græsbøll Christensen and Prof. Richard Heusdens (NLDA).

Audio Analysis Lab Ph.D. student Charlotte Sørensen succesfully defended her thesis entitled “Non-Intrusive Speech Intelligibility Prediction” on Monday August 24 2020. Her thesis was supervised by Jesper Bünsow Boldt and Mads Græsbøll Christensen in a collaboration between AAU and GN Advanced Science. The assessment committee was comprised of distinguished researchers Sofia Dahl, Tiago Falk and Tao Zhang.

The IEEE’s flagship conference within signal processing ICASSP 2020 was held May 4-8 2020, and it was supposed to be in beautiful Barcelona in Spain. However, due to Covid-19, the conference was all virtual this year with videos of all presentations. This year, Audio Analysis Lab members presented the following papers:

• Title: CONVEX OPTIMISATION-BASED PRIVACY-PRESERVING DISTRIBUTED AVERAGE CONSENSUS IN WIRELESS SENSOR NETWORKS Authors: Qiongxiu Li, Richard Heusdens, Mads Græsbøll Christensen
• Title: ROBUST FUNDAMENTAL FREQUENCY ESTIMATION IN COLOURED NOISE Authors: Alfredo Esquivel Jaramillo, Andreas Jakobsson, Jesper Kjær Nielsen, Mads Græsbøll Christensen
• Title: A FAST REDUCED-RANK SOUND ZONE CONTROL ALGORITHM USING THE CONJUGATE GRADIENT METHOD Authors: Liming Shi, Taewoong Lee, Lijun Zhang, Jesper Kjær Nielsen, Mads Græsbøll Christensen
• Title: AUTOREGRESSIVE PARAMETER ESTIMATION WITH DNN-BASED PRE-PROCESSING Authors: Zihao Cui, Changchun Bao, Jesper Kjær Nielsen, Mads Græsbøll Christensen

Aside from these presentations, the Audio Analysis Lab also presented the following demo:

• Title: User Tuneable Sound Zones Authors: Taewoong Lee, Liming Shi, Jesper Kjær Nielsen, Mads Græsbøll Christensen

All talks are available on the ICASSP 2020 virtual platform and on the Audio Analysis Lab’s YouTube Channel.

On January 21 2020, Audio Analysis Lab member Liming Shi sucecssfully defended his PhD thesis entitled Speech Modeling and Robust Estimation for Diagnosis of Parkinson’s Disease. The thesis is concerned with new models and methods for diagnosing Parkinson’s disease from voice signals recorded under noisy conditions. The assessment committee was comprised of Prof. Jonathon Chambers (University of Leicester), Dr. Phil Garner (IDIAP), and Assoc. Prof. Troels Pedersen (AAU). Liming Shi was supervised by Prof. Mads Græsbøll Christensen and co-supervised by Assoc. Prof. Jesper Rindom Jensen and Assoc. Prof. Jesper Kjær Nielsen.