CBFdataset: A Dataset of Chinese Bamboo Flute Performances

<p><em>CBFdataset</em>&nbsp;is<em>&nbsp;</em>a dataset of Chinese bamboo flute (CBF) performances, created for ecologically valid analysis of music playing techniques in context.</p> <p>The dataset comprises monophonic recordings of classic CBF pieces and isolated playing techniques, recorded by 10 professional CBF performers; and expert annotations of seven playing techniques: vibrato, tremolo, trill, flutter-tongue (FT), acciaccatura, portamento, and glissando. The recorded pieces include&nbsp;<em>Busy Delivering Harvest (BH)</em> 扬鞭催马运粮忙, <em>Jolly Meeting (JM)</em> 喜相逢, <em>Morning (Mo)</em> 早晨, <em>and Flying Partridge (FP)</em> 鹧鸪飞.&nbsp;All data was recorded in a professional recording studio using a Zoom H6 recorder at 44.1kHz/24-bits. The difference between different Versions 1.2, 1.1, and 1.0:</p> <ul> <li>V1.2 is the complete CBFdataset&nbsp;with a total duration of 2.6 hours.</li> <li>V1.1 splits the CBFdataset into two subsets according to playing technique types: CBF-periDB&nbsp;and CBF-petsDB. The former contains all the&nbsp;full-length pieces, isolated playing techniques, and annotations of four periodic modulations: vibrato, tremolo, trill, and flutter-tongue. The latter comprises the same full-length recordings, isolated playing techniques, and annotations of three pitch evolution-based techniques: acciaccatura, portamento, and glissando.</li> <li>V1.0 includes only the CBF-periDB.</li> </ul> <p>Related&nbsp;updates, demos, and code for reproducibility are available at&nbsp;<a href="http://c4dm.eecs.qmul.ac.uk/CBFdataset.html">http://c4dm.eecs.qmul.ac.uk/CBFdataset.html</a>.&nbsp;Any queries, please feel free to contact Changhong at&nbsp;changhong.wang@telecom-paris.fr. Please cite the following paper when using this dataset:</p> <p>Changhong Wang, Emmanouil Benetos, Vincent Lostanlen, and Elaine Chew,&nbsp;&quot;Adaptive Scattering Transforms for Playing Technique Recognition,&quot;&nbsp;<em>IEEE/ACM Transactions on Audio, Speech, and Language Processing (TASLP)</em>, 30 (2022): 1407-1421.</p>