Dynamic Percussion Dataset

Note: latest version (1.1.0) here

This dataset was recorded as a part of my Bachelor's thesis work for Tampere University. It is a collection of samples from different drums and percussion, recorded in an anechoic room. More details on the structure, labels and recording process can be found in [1].

In the thesis, this dataset was visualized in the 2-dimensional space with a method (described in [1]) that utilizes OpenL3 deep audio embeddings [2] and t-SNE dimensionality reduction algorithm [3]. Since the embeddings can take a significant time to compute without a sufficient CUDA environment installed, they are provided with the dataset (in a later version).

Audio format:
Data type: wav
Sampling rate: 44,1 kHz
Bit-depth: 24
Channels: 1

NOTE: The reference [1] will be added when the thesis has been published.