Meshal Alokifi

تخصص الذكاء الاصطناعي · دفعة 2023

Qassid

Qassid can generate Saudi poetry by giving it input text to begin with, and then complete the poem based on the input. Also, Qassid is able to predict the era in which the poetry was written. To overcome this challenge, we trained an LSTM model on labeled speech samples, enabling it to adeptly capture speech dialect patterns and classify new samples into their respective categories. Throughout our experimentation process, we discovered that oversampling the underrepresented dialects in our dataset was the most effective approach.”

اطّلع عليه في تقرير البرنامج (PDF)