CHAD-Hummings Subset

This repository contains the hummings subset of the dataset from "A Semi-Supervised Deep Learning Approach to Dataset Collection for Query-by-Humming Task" (ISMIR 2023).

For the complete dataset and further details, please visit the main GitHub repository.

Overview

The chad_hummings_subset.tar.gz archive provided in this repository contains a collection of 5,314 humming audio files.

These audio files are sorted into groups of 693 distinct humming fragments originating from 311 unique songs (groups).

Audio format - .wav.

Dataset Structure

Upon extracting the dataset from chad_hummings_subset.tar.gz, you will find the following structured hierarchy:

├── {GROUP_ID}             
│   ├── {FRAGMENT_ID}        
│       ├── {ID}.wav
│       └── ...
│   └── ...
└── ...

where

GROUP_ID - the unique identifier for each song,
FRAGMENT_ID - the identifier for individual fragments within a song,
ID - the version identifier for a specific fragment of the song.

This structured hierarchy organizes the audio files and fragments, making it easier to navigate and work with the dataset.

Citation

Please cite the following paper if you use the code or dataset provided in this repository.

@inproceedings{Amatov2023,
    title={A Semi-Supervised Deep Learning Approach to Dataset Collection for Query-by-Humming Task},
    author={Amatov, Amantur and Lamanov, Dmitry and Titov, Maksim and Vovk, Ivan and Makarov, Ilya and Kudinov, Mikhail},
    year={2023},
}