Dataset for file fragment classification of audio file formats

Area of Science:

• Digital Forensics
• Computer Science
• Signal Processing

Background:

• File fragment classification is vital for network forensics.
• Existing public datasets for audio file formats are limited.
• A lack of standardized datasets hinders performance comparison of classification methods.

Purpose of the Study:

• To introduce a novel dataset for audio file fragment classification.
• To facilitate reproducible research and performance benchmarking in network forensics.
• To support the development and evaluation of audio file format identification techniques.

Main Methods:

• The study presents a dataset comprising file fragments from 20 distinct audio file formats.
• Each format includes fragments from audio files with varied compression settings.
• The dataset contains a total of 20,160 file fragments, with 210 for each format-compression pair.

Main Results:

• A comprehensive dataset of audio file fragments has been created.
• The dataset covers 20 common and specialized audio file formats.
• It provides a standardized resource for evaluating file fragment classification algorithms.

Conclusions:

• The developed dataset fills a critical gap in public resources for audio file forensics.
• It will enable more reliable comparisons of classification methods.
• This resource is expected to advance the field of network forensics and digital audio analysis.