Skip to content

RapBank

A rap music dataset with a preprocessing pipeline

Visit official site

What RapBank is

The dataset is described as a rap music dataset collected from YouTube, supplied with a preprocessing pipeline for training and testing music generation models.

What you can do with it

  • Analyse rap styles with the collected songs
  • Train a rap music generation model
  • Use the collection as teaching material

Getting started

  • Read the details and terms on the GitHub page
  • Download the dataset into a local path
  • Run pipeline.sh with input and feature output paths

What to watch out for

  • Total duration is stated as 5,586 hours
  • Average song length is stated as 218 seconds
  • English songs take about two thirds of total duration

Pros & cons

✓ What we like

  • A pipeline covers source separation, splitting and lyric recognition
  • The dataset suits model training and testing

! What to watch out for

  • Downloads are stated for non-commercial use
  • Use follows the CC BY-NC-SA 4.0 licence

FAQ

How large is the dataset?

The material lists 94,164 links, 92,371 downloaded songs and 5,586 hours.

Can it be used commercially?

The feature list states non-commercial use under CC BY-NC-SA 4.0.

Last reviewed: 2026-11-15

More AI music creation tools

View all →

How we review