#If you use this dataset please cite our paper
Link to the paper: https://rdcu.be/dIJX0
You can download the datasets from the folowing links:
English Dataset:
https://drive.google.com/file/d/1XgO4F39fl17qkcvci0iMglDaUo4-VvDa/view?usp=sharing
Arabic Dataset:
https://drive.google.com/file/d/1x7wxxdRvrBTI5fsjZ8M0BCxe5-_eunnN/view?usp=sharing
Mixed Dataset CSV (download both and compine them as one mixed dataset):
https://drive.google.com/file/d/1x7wxxdRvrBTI5fsjZ8M0BCxe5-_eunnN/view?usp=sharing
here are 2 datasets of audios, the first one is the the English dataset which is all the files expect the arabic_fake and arabic_real to create the first dataset and use it download all the audios and compine them in a file then use 90% of these audios as a training set (10% of it is validation which the model take without the need to do it manually) while the remaining 10% for testing
the Arbic dataset is by compining the files with the English that you already have to create the mixed dataset and as before 90% is for training (10% of it is validation which the model take without the need to do it manually) and 10% for testing
NOTE: -to know the real and fake audios during tarining the audios that ends with 'r' are real and the one ands with 'f' are faked. -The format of all the audios is .WAV -the duration of the audios is 20 second long