sireNNet-Emergency Vehicle Siren Classification Dataset For Urban Applications

Published: 14 February 2023| Version 1 | DOI: 10.17632/j4ydzzv4kb.1
Contributors:
Arya Shah,

Description

An audio dataset for classifying sirens of emergency vehicles such as firetruck, ambulance and police vehicles along with an additional class for traffic noise. The dataset consists of the following 4 classes with the distribution of data within them as follows: 1. Ambulance: 400 files 2. Police: 454 files 3. Firetruck: 400 files 4. Traffic: 421 files Since the Dataset has gone through multiple iterations, the authors have uploaded the final version ONLY which includes the augmented datapoints. Additionally, the extracted features of the original dataset, augmented dataset, code files, notebooks and subsequent research paper can be availed on-demand by approaching the authors personally.

Files

Steps to reproduce

Steps followed for collection of dataset is as follows: 1. Scrape vehicle siren audios from internet 2. Sample the .wav audio files 3. Normalize the sound files 4. Augment the dataset 5. Additionally, preprocess the audio files if needed, in terms of audio enhancement, noise reduction, etc.

Institutions

Narsee Monjee Institute of Management Studies University, Mukesh Patel School of Technology Management and Engineering

Categories

Machine Learning, Audio Analysis, Convolutional Neural Network, Deep Learning, Long Short-Term Memory Network

Licence