pollitoconpapass/dataset-peru-shazam

Peru Shazam Dataset

A collection of Peruvian music tracks discovered via Shazam, covering traditional and popular genres from Peru.

Dataset Summary

Property	Value
Total tracks	318
Audio format	WAV (44.1kHz mono)
Genres	5
Source	YouTube (Shazam discovery)

Genre Distribution

• chicha: 96 tracks
• criolla: 13 tracks
• cumbia_amazonica: 91 tracks
• huayno: 18 tracks
• rock: 100 tracks

Data Structure

dataset-peru-shazam/
├── README.md          # This file
├── catalog.json       # Full metadata (title, genre, duration, source URLs)
└── audio/             # WAV audio files (named by YouTube ID)

catalog.json Fields

Field	Type	Description
id	string	YouTube video ID (used as filename)
title	string	Song title / description
duration_seconds	int	Track duration in seconds
genre	string	Music genre classification
filename	string	Audio filename
source_url	string	Original YouTube URL

Usage

As the catalog.json is important to check the audio metadata, this is the ideal way to use this dataset.

Use of the Audio Data

from datasets import load_dataset

ds = load_dataset("pollitoconpapass/dataset-peru-shazam")

sample = ds[0]['audio'] # -> This is the audio itself of the first item: <datasets.features._torchcodec.AudioDecoder object at 0x7ef124d88150>

You can use the ds variable as the audios dataset

Use of the JSON catalog

from datasets import load_dataset

catalog_json = load_dataset("pollitoconpapass/dataset-peru-shazam", data_files="catalog.json")
# When you apply data_files you can only use the file attached there...

json_content = catalog_json['train'][0] # -> this is the whole JSON object in catalog.json

print(json_content['R0V41zBH2xY'])
# Output of line above:
'''
{'id': 'R0V41zBH2xY',
 'title': 'Que linda flor - Silverio Urbina',
 'duration_seconds': 252,
 'genre': 'huayno',
 'filename': 'R0V41zBH2xY.wav',
 'source_url': 'https://www.youtube.com/watch?v=R0V41zBH2xY',
 'status': 'downloaded',
 'discovered_at': '2026-09-22T16:04:13.743392+00:00',
 'downloaded_at': '2026-09-22T17:11:16.063261+00:00',
 'error': None}
'''

You can use the json_content variable as the guide for all the audios present in the dataset

License

CC-BY-NC-4.0 — For non-commercial research and educational use.