Vehicle horn sound classification datasets

Description of the data and file structure

Car Horn Sounds Dataset

Overview

This dataset comprises a collection of car horn sounds extracted from the popular video game Forza Horizon 4. It includes recordings of 10 distinct types of car horns, meticulously captured and segmented into individual WAV files for ease of use in various audio processing, machine learning, and research applications.The simulated dataset consists of 800 audio signals for each different vehicle model, divided into four groups: 200 horn audio samples without external noise, 200 samples overlaid with white noise, 200 samples accompanied by simulated wind and rain sounds, and 200 samples with controlled pitch and speed modifications. Each honk audio sample in the dataset maintains a consistent duration of 5 seconds.

Table of Contents

• Overview
• Dataset Contents
• Data Collection Process
• Usage
• Acknowledgements
• Copyright
• Disclaimer

Dataset Contents

• Total Categories: 10 different car horn types
• File Format: WAV
• Number of Files: Multiple files per horn type, totaling [200] recordings
• Duration: Each recording ranges from [5 seconds]

Data Collection Process

The dataset was created using Audacity, a free, open-source audio editing software, in conjunction with the video game Forza Horizon 4. The process involved the following steps:

1. Environment Setup:
2. • Launched Forza Horizon 4 and navigated to scenarios featuring various car models equipped with different horn sounds.
   • Configured Audacity to capture high-quality audio directly from the system's sound output.
3. Recording:
4. • Played the game and triggered each of the 10 distinct car horn types multiple times to ensure a diverse set of recordings.
   • Each horn sound was recorded in real-time, capturing the authentic audio output from the game environment.
5. Segmentation:
6. • Imported the raw recordings into Audacity.
   • Utilized Audacity’s editing tools to accurately split the recordings into individual WAV files, each containing a single horn sound.
   • Ensured that each segment was free from background noise and artifacts, maintaining high audio quality.
7. Organization:
8. • Categorized the WAV files based on horn type, labeling them systematically for easy identification and access.
   • Verified the integrity of each file to ensure consistency and reliability across the dataset.

Usage

This dataset is ideal for a variety of applications, including but not limited to:

• Machine Learning: Training models for audio classification, recognition, and synthesis.
• Audio Processing: Testing and developing sound editing and enhancement algorithms.
• Research: Conducting studies on audio perception, sound differentiation, and virtual sound environments.

Acknowledgements

Special thanks to the developers of Audacity and Forza Horizon 4 for providing the tools and environments necessary to create this dataset. Your contributions to audio software and gaming have made this project possible.

Copyright

The audio content in this dataset is sourced from the Forza Horizon 4 game, and all rights are reserved by the original developers. Before using and distributing these audio files, please ensure you have obtained the necessary permissions or licenses. Unauthorized use may violate copyright laws.

Disclaimer

This dataset is shared in accordance with the data availability requirements of the PLOS ONE journal. The audio files are extracted from Forza Horizon 4, and their distribution is intended solely for research and educational purposes. By providing this dataset, we acknowledge that the audio content is the property of Playground Games and Microsoft Studios. Users are responsible for ensuring compliance with all applicable copyright laws and obtaining any necessary permissions for use beyond the scope of this dataset.

If you believe that any content in this dataset infringes upon your copyright, please contact kangxii@foxmail.com to request its removal.

If you have any questions or need further assistance regarding this dataset, feel free to contact kangxii@foxmail.com.

Access information

Data was derived from the following sources:

• Forza Horizon 4 by Playground Games and Microsoft Studios: The audio recordings were extracted directly from this video game.
• Audacity: Used for recording and segmenting the audio files.