Published November 1, 2024 | Version v1
Dataset Open

Sounds of Vehicle Internal Combustion Engines

  • 1. University of Niš, Faculty of Electronic Engineering, Serbia
  • 2. Academy of Technical and Art Applied Studies, School of Electrical and Computer Engineering, Serbia

Description

Database Description

By applying a specially developed acquisition system at the entrance to an underground garage, a total of 350 cars were sampled over four iterations. Based on the collected samples, a database containing 475 sound samples was formed.

In the first three iterations, different approaches were applied regarding sample duration. Within all three approaches, one or two audio recordings per vehicle were obtained for most vehicles, where the first recording consistently represented the stationary engine idle mode without exception. In most samples, the second recording (in some cases the last one) partially contained the engine idle mode followed by an increase in crankshaft speed and a partial engine load mode for the purpose of vehicle acceleration. There were no cases in which the partial engine load mode appeared in the recordings before the idle mode.

By applying an algorithm for extracting the idle mode from sound samples that, in addition to idle mode, also contained a change in operating mode, the moments up to which the signal remained stationary were determined. A smaller number of recordings were discarded as too short, while some recordings were shortened by extracting only the idle mode segment.

After applying the idle mode extraction algorithm, which also eliminated electric vehicles, the final number of usable samples was 320.

Files

Diesel Motor Sounds.zip

Files (119.2 MB)

Name Size Download all
md5:962e7d025e0efbfcfa1e56960c85c5b2
68.1 MB Preview Download
md5:896accd703c04b46af23485698ce6c45
51.1 MB Preview Download

Additional details

Related works

Is described by
Journal article: 10.1109/JSEN.2024.3516876 (DOI)
Journal article: 10.1016/j.heliyon.2025.e43426 (DOI)

Funding

European Commission
AIDA4Edge - Twinning for Excellence in Adaptive Edge AI 101160293
The Slovenian Research and Innovation Agency
P2-0401
The Slovenian Research and Innovation Agency
J7- 50042

Dates

Updated
2024-11-01