Overview

Query by tapping (QBT) is the process of identifying a musical excerpt based upon a tapped representation. QBT is a canonical Music Information Retrieval (MIR) task and an intuitive query to perform, yet the literature on this topic remains small relative to other query forms such as singing and humming.

QBT-Extended is a dataset for query by tapping. For further information, please read our ISMIR paper.

If you publish results obtained using this dataset, or reference it in anyway, please cite as follows:
Blair Kaneshiro, Hyung-Suk Kim, Jorge Herrera, Jieun Oh, Jonathan Berger and Malcolm Slaney. QBT-Extended: An Annotated Dataset of Melodically Contoured Tapped Queries. In Proceedings of the 14th International Society for Music Information Retrieval Conference, Curitiba, Brazil, November, 2013.

Erratum

Some results published in the ISMIR 2013 paper have changed slightly. More details can be found in the code page.

Dataset

A description of the dataset can be found here.

Download

You can get the complete dataset (includes SQLite3, CSV and .onset files, as well as the ground truth data).

The dataset is also provided in three different formats separately:

  1. SQLite3 database
  2. CSV files
  3. Onset files
You can also get the ground truth data (MIDI renditions of the excerpts included in the dataset, in MIDI and .onset formats).

Creative Commons License
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 3.0 United States License.

Data collection system

This dataset was collected using an iOS user interface and a Ruby-on-Rails backend using an SQLite3 database. The complete source code can be accessed here.