WaivOps Lo-Fi Chords Dataset: Open Audio Resources for Machine Learning in Music
Creators
Description
Lo-Fi Chords Dataset
Lo-Fi Chords Dataset is an open audio collection of chord progressions in the style of retro soul and library production music. It features 8,000 audio loops in uncompressed stereo WAV format, along with paired JSON files containing metadata for training generative AI audio models.
Overview
The dataset was created using custom scripting to render audio from a database of MIDI patterns and multi-sample instruments. Recordings feature flattened chord progressions generated with various acoustic, electric, and synthesizer piano soundfonts. Data augmentation included humanization, time shifting, filtering and equalization, tape/vinyl noise, and convolution reverb modeling. These strategies enhance model generalization by exposing training examples to varied piano qualities and ambient effects.
Its primary purpose is to provide accessible content for model development, audio research, and music applications. Example uses include text-to-audio, style transfer, feature extraction, tempo detection, audio classification, rhythm analysis, music information retrieval (MIR), sound design, and signal processing.
Specifications
- 8,000 8-bar audio loops (~54 hours)
- 16-bit stereo WAV format, 44.1 kHz
- Tempo range: 60–100 BPM
- Paired metadata files (WAV + JSON)
- Instruments: Grand piano, Upright piano, Electric piano, Pianet, FM synth piano
- Subgenre styles: Soul, Retro library music
License
This dataset was compiled by WaivOps, a crowdsourced music project managed by Patchbanks. All recordings have been sourced from verified composers and providers for copyright clearance.
The Lo-Fi Chords Dataset is licensed under Creative Commons Attribution 4.0 International (CC BY 4.0).
Additional Info
For audio examples or more information about this dataset, please refer to the GitHub repository.