MICM Music Dataset

The MICM (Maryam Iranian Classical Music Dataset) Speech Dataset is a music dataset for classification and generation with Deep learning techniques. The given dataset includes 1137 music samples which contains 631 music samples with the foreground of Ney instrument sound and also some other instruments in the background. The remained music samples include the Violin instrument sound as the foreground sound. Aiming to present an instrument Independent method to classifying and generating Dastgah musics, two musical instruments are employed, Violin and Ney.
The classification process of 7 Dastgahs of Iranian classical music on this dataset has shown best results ever .
Contents
The corpus is downloadable from its Kaggle web page, and contains the following:
* The MICM dataset contains seven classes which provides several music samples in seven Iranian Classical Dastgahs namely as follows: Shour, Homayoun Mahour, Segah, Chahargah, Rastpanjgah and Nava. Each music samples have different numbers of signal samples and also the sample rate of each music sample is 2318.
* In kaggle web page, Data samples (.wav files) are in 'Data/Music/SoundSamples' directory. 'Music' Directory contains 2 directories 'Avaz' and 'Dastgah' each of which contains 2 '.txt' files. 'Class.txt' contains the name and number of each class and 'Label.txt' contains class number of each sample '.wav' file.

Main Menu

See Also

Main Menu