Mridangam Stroke Dataset

The Mridangam Stroke dataset is a collection of 7162 audio examples of individual strokes of the Mridangam in various tonics. The dataset comprises of 10 different strokes played on Mridangams with 6 different tonic values.

The dataset, showing the number of examples in each category: different stroke labels shown in rows and different tonic values shown in columns
	B	C	C#	D	D#	E	Total
Bheem	5	3	1	0	15	25	49
Cha	57	50	54	67	49	53	330
Dheem	127	86	78	12	111	54	468
Dhin	48	48	63	12	198	113	482
Num	81	98	97	18	143	60	497
Ta	145	165	217	180	119	105	931
Tha	200	185	211	224	196	160	1176
Tham	88	80	35	29	92	50	374
Thi	438	334	369	283	444	345	2213
Thom	136	80	72	91	128	135	642
Total	1325	1129	1197	916	1495	1100	7162

Download

This dataset can be downloaded here.

Each audio file is named as,

<StrokeName>_<Tonic>_<InstanceNumber>.wav
<Tonic> = {B, C, Csh, D, Dsh, E}
<StrokeName> = {Bheem, Cha, Dheem, Dhin, Num, Ta, Tha, Tham, Thi, Thom}

Audio content

The dataset provides audio examples for each of the strokes. The number of examples is shown in Table 1, for six different tonics and ten different stroke labels.

The audio examples were recorded from a professional Carnatic percussionist in a semi-anechoic studio conditions by Akshay Anantapadmanabhan using SM-58 microphones and an H4n ZOOM recorder. The audio was sampled at 44.1 kHz and stored as 16 bit wav files. The dataset can be used for training models for each Mridangam stroke.

Reference

A detailed description of the Mridangam and its strokes can be found in the paper below. A part of the dataset was used in the following paper. Please cite it if you use the dataset in your work:

Akshay Anantapadmanabhan, Ashwin Bellur, Hema A. Murthy, "Modal analysis and transcription of strokes of the mridangam using non-negative matrix factorization," in Proc. of the IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP 2013), pp.181-185, May 2013

If you have any questions or comments about the dataset, please feel free to write to us:

Akshay Anantapadmanabhan ()