Psychoacoustics and Machine Learning Methods for the Objective Rating of Creep Groan Noise

  • Huemer-Kals, S. (Speaker)
  • Máté Tóth (Contributor)
  • Jurij Prezelj (Contributor)
  • Martin Zacharczuk (Contributor)
  • Karl Häsler (Contributor)
  • Fischer, P. (Contributor)

Activity: Talk or presentationTalk at conference or symposiumScience to science

Description

Rating a brake and axle setup’s creep groan performance is a necessity during the development process of a vehicle, be it for the comparison of different components or for the final assessment and release. Objective rating methods present a valuable addition to the often laborious and biased subjective ratings by test drivers.

One solid approach for objective rating can be based on psychoacoustic metrics: These metrics quantify specific auditory sensations of the human being. Based on preliminary evaluations presented at EuroBrake 2021, the psychoacoustic metrics loudness, sharpness, roughness, fluctuation strength and tonality were analysed in detail, both with classical statistical methods as well as with machine learning approaches.

Despite a certain variance in the subjective rating data, it was possible to create accurate rating models. Depending on the used signal source, cabin sound pressure or brake caliper acceleration, the metrics loudness, roughness, and tonality were found to dominate the human sensation of creep groan. Furthermore, the behaviour of different creep groan manifestations was compared. Based on the presented results, accurate objective rating mechanisms can be developed and implemented in future brake NVH development processes.
Period19 May 2022
Event titleEuroBrake 2022
Event typeConference
LocationVirtuell, United KingdomShow on map
Degree of RecognitionInternational

Fields of Expertise

  • Mobility & Production