CBFdataset: A Dataset of Chinese Bamboo Flute Performances

CBFdataset is a dataset of Chinese bamboo flute (CBF) performances, created for ecologically valid analysis of music playing techniques in context. The dataset contains monophonic recordings of classic CBF pieces and expert playing technique annotations. It includes two subsets:

• CBF-periDB [1]: a subset for periodic modulation analysis in music signals. It comprises full-length pieces and annotations of four types of periodic modulations: vibratos, tremolos, trills, and flutter-tongue. 
• CBF-petsDB [2]: a subset for analysing pitch evolution-based techniques (PETs). It includes full-length pieces and annotations of three types of PETs: acciaccatura, portamento, and glissando.

Relevant updates, demos, and code for reproducibility are available at http://c4dm.eecs.qmul.ac.uk/CBFdataset.html. Any queries, please feel free to contact Changhong at changhong.wang@qmul.ac.uk. Please cite the corresponding papers:

[1] C. Wang, E. Benetos, V. Lostanlen, and E. Chew, "Adaptive Time–Frequency Scattering for Periodic Modulation Recognition in Music Signals," In Proceedings of the International Society for Music Information Retrieval Conference (ISMIR), pages 809–815, 2019.

[2] C. Wang, V. Lostanlen, E. Benetos, and E. Chew, "Playing Technique Recognition by Joint Time–Frequency Scattering". In Proceedings of the IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pages 881–885, 2020.