QSDAS : new Quranic speech database for Arabic speaker recognition

الملخص EN

-A speech database is the foundation for developing any speaker processing system or algorithm.

To cope with various requirements to speaker recognition technology, the Quranic Speech Database for Arabic Speakers (QSDAS) has been produced.

This database is a useful resource to aid in the assessment of Arabic speaker recognition and linguistic research.

It consists of 15.4 GB representing 6489 files, including 1617 files (Surahs) read aloud by 77 male speakers.

The 1617 audio files (6.4 GB) are partitioned in three sets: SetD, SetS, and SetT with 77, 770, and 770 audio files, respectively.

SetD contains one file per speaker used for development, while SetS and SetT contain ten files per speaker and are used for tests purposes.

In addition to the 1617 audio files, QSDAS includes for each audio file the corresponding fundamental frequency F0, F1-F2-F3 formants file, and MFCC and Energy file to make it usable immediately for speaker recognition research.