Fraunhofer IDMT Ilmenau presents a broad range of scientific contributions at the DAGA 2023 conference for acoustics

March 02, 2023

From March 6 to 9, 2023, the 49th Annual Conference on Acoustics - DAGA 2023 - will take place in the northern German metropolis of Hamburg. This year, the Ilmenau-based Fraunhofer IDMT is participating in the symposium with a total of seven presentations, two scientific posters, and the chairing of a structured session. The scientific contributions will present current research topics of the Ilmenau institute in the fields of sound analysis for music and acoustic events, numerical acoustics, acoustic metamaterials as well as electroacoustic, and audio signal processing.

Institute director Prof. Joachim Bös, who has been a member of the DAGA board since July 2022, is pleased that Fraunhofer IDMT will be represented with a multifaceted lecture program: "This year, my colleagues from the main site in Ilmenau, but also from the Oldenburg branch of the institute, will illuminate many different aspects of acoustics in their contributions and exhibits. Above all, the transfer of scientific findings into practical applications plays an essential role."

The following is a brief overview of the topics that Ilmenau scientists will present at this year's DAGA.

Last modified: March 02, 2023

Acoustic precipitation measurement

Oral Presentation: »Investigations on the implementation of an acoustic rain sensor system« | Kevin Hock | March 8, 2023 | 9:20 a.m. | room Y 8

Picture shows an experimental setup at Fraunhofer IDMT for generating rain test data. — © Kevin Hock/Fraunhofer IDMT
Experimental setup at Fraunhofer IDMT for generating rain test data.

Fraunhofer IDMT is researching an acoustic rain sensor to measure precipitation accurately in terms of time and course. Unlike the conventional method, in which rain is collected in a defined container and the amount of water is read off after a certain time, the acoustic rain sensor can be used to collect data on the course of precipitation with high temporal resolution. The first results of scientific experiments will be presented in the lecture.

Abstract

Bioinspired microphones

Oral Presentation: »Influence of Sensor Design on Bio-inspired Adaptive Acoustic Sensing« | Daniel Beer | March 9, 2023 | 9:00 a.m. | room X 3 – 4

Schematic representation of a future bioinspired microphone. — © Fraunhofer IDMT
Schematic representation of a bioinspired microphone with an overall size of less than three millimeters.

Together with the Ilmenau University of Technology, Fraunhofer IDMT is working on a new principle of bioinspired microphones. The work on this is presented in a paper. The idea: Based on the biological hearing process, the signal processing and the dynamic adaptation of the detection and processing characteristics are integrated into a sensor stage. The idea is that when used in hearing aids, people can hear better with it in noisy environments.

Abstract

Acoustically optimized metamaterials

Oral Presentation: »Designing Active Acoustic Metamaterials« | Georg Fischer | March 8, 2023 | 6:20 p.m. | room Y 8

Acoustic metamaterials can be used as sound insulation measures in a wide range of industries, such as mechanical engineering or vehicle construction. Thanks to their properties, acoustic metamaterials reduce disturbing airborne noise. In the lecture, an approach to the development of metamaterials will be presented that combines methods from acoustics and electronics.

Abstract

Acoustic data acquisition instead of optical monitoring

Poster Presentation: »Acoustic data acquisition for quality monitoring during Powder Bed Fusion with Laser Beam (PBF-LB)« | Yaxiong Ren (TU Darmstadt, Fachgebiet SAM) | March 8, 2023 | 3:40 pm | room Foyer Z, poster section M

Picture shows the production of a component by additive manufacturing — © adobestock.com/phuchit
To monitor additive manufacturing processes even more reliably, Fraunhofer IDMT is developing methods for acoustic monitoring of manufacturing sounds.

On the basis of a scientific poster, current project results in the field of acoustic data acquisition for quality monitoring in powder bed fusion with laser beam will be presented. With this additive manufacturing process, components are built up by applying and fusing metal powder layer by layer. Process monitoring is currently done with optical methods, but this only reveals superficial defects. Fraunhofer IDMT will present a method to detect manufacturing defects using ultrasonic microphones.

Abstract

Simulated sound data

Oral Presentation: »Development of Machine Learning Algorithms Using Virtual Sonic Data in Lightweight Construction for Quality Assurance in Additive Manufacturing (ML-S-LeAF)« | Ömer Faruk Yildiz | March 8, 2023 | 4:40 p.m. | room X 5 – 6

In connection with the acoustic monitoring of manufacturing processes, such as additive manufacturing, another scientific presentation will focus on virtual sonic data. The aim is to use not only "real" measurement data, but also numerical sound data obtained from acoustic simulations for training machine learning algorithms. The presentation will preview intermediate results from first Machine Learning experiments and provide a comparison between measured and simulated sonic data.

Abstract

Secure and trustworthy acoustic data analysis

Oral Presentation: »Is your data safe? Privacy and Trustworthiness aspects of Neural Network based Acoustic Analysis» | Thomas Köllmer | March 9, 2023 | 10:40 a.m. | room X 3 – 4

Immediately with the acquisition and analysis of acoustic and audio data, the question of the trustworthiness of the processes and the privacy of the data becomes more and more pressing - especially when data are analyzed cloud-based and with Deep Learning methods. In the lecture, current issues and questions will be explained and technical approaches will be shown.

Abstract

Raspberry Pi-based versus professional speaker measurement systems.

Poster Presentation: »Measurement of Loudspeaker Parameters with a Raspberry Pi« | Juan Manuel Rodriguez Mejia | March 9, 2023 | 11:00 a.m. | room Foyer Z, poster section J

Bild shows Raspberry Pi-based measurement system for analyzing the parameters of a loudspeaker. — © Juan Manuel Rodríguez Mejía
Raspberry Pi-based measurement system for analyzing the parameters of a loudspeaker.

Raspberry Pis are full-fledged computers in miniature format, with all necessary components built on a single circuit board. Due to their compact design and powerful performance, Raspberry Pis are well-suited for implementation in measurement systems. In a poster session, the results of a series of experiments comparing Raspberry Pi-based measurements with a professional loudspeaker measurement in terms of reliability and accuracy will be presented.

Abstract

Structured Session "Sound Analysis for Music and Audio Signals" led by Fraunhofer IDMT

Session Management: Dr.-Ing. Jakob Abeßer

On March 7 and 8, 2023, the structured session "Sound Analysis for Music and Audio Signals" will present Deep Learning approaches for analysis, segmentation, and classification of audio signals. The advantages and limitations of such "black boxes" will be discussed and, in comparison, explainable Deep Learning models capable of better understanding structural and acoustic properties of sound events will be presented. Our expert for intelligent semantic audio signal processing, Dr.-Ing. Jakob Abeßer is leading the session together with his scientific colleagues Prof. Meinard Müller (AudioLabs Erlangen) and Prof. Dr.-Ing. Sebastian Stober (Otto-von-Guericke University Magdeburg).

Classification of music genres and training of machine learning models for sound and music analysis

The structured session will also feature two presentations by researchers from Fraunhofer IDMT Ilmenau:

Oral Presentation: »Quantifying Uncertainty in Music Genre Classification« | Hanna Lukashevich | March 8, 2023 | 9:20 a.m. | room Y 8

This talk discusses issues and findings around the robust classification of music genres.

Abstract

Oral Presentation: »An Introduction to Unsupervised Domain Adaptation in Sound and Music Processing« | Franca Bittner | March 7, 2023 | 5:20 p.m. | room Y 8

In this presentation, the researchers speak about the "Unsupervised Domain Adaptation" approach to improve the robustness and performance of Machine Learning models for sound and music processing.

Abstract

Fraunhofer IDMT Ilmenau at the DAGA 2023 conference for acoustics

Fraunhofer IDMT Ilmenau presents a broad range of scientific contributions at the DAGA 2023 conference for acoustics

Acoustic precipitation measurement

Bioinspired microphones

Acoustically optimized metamaterials

Acoustic data acquisition instead of optical monitoring

Simulated sound data

Secure and trustworthy acoustic data analysis

Raspberry Pi-based versus professional speaker measurement systems.

Structured Session "Sound Analysis for Music and Audio Signals" led by Fraunhofer IDMT

Classification of music genres and training of machine learning models for sound and music analysis

You might also be interested in

The DAGA website with program

Acoustics at Fraunhofer IDMT Ilmenau

Acoustic monitoring at Fraunhofer IDMT Ilmenau

Audiovisual analysis at Fraunhofer IDMT

Contact Press / Media

Julia Hallebach