Pronunciation Errors in Arabic YouTube Videos Narrated by AI

Arabic has three long vowels /a:/, /u:/, /i:/ and three short vowel /a/, /u/, /i/ which are represented by diacritics marked over and under consonant letters. In words that have short vowels, only the consonants are written. Arabs usually read without any diacritics except for the Holy Quran and the Prophet’s Hadiths. The absence of short vowel diacritics poses pronunciation difficulties for Artificial Intelligence (AI).  Henceforth, this study aims to anlayze a sample of Arabic YouTube videos narrated by AI to find out which Arabic words are mispronounced by AI narrators. It was noted that AI narrators speak with a natural voice, good expression and intonation. They make no grammatical or syntactic errors. But they make pronunciation errors, especially in diacritics and homographs (words that are spelled the same but have different pronunciations and meanings depending on the short vowel diacritics which are not usually shown on written words). This means that AI has difficulty matching the pronunciation of a homophone with the context in which it is used. They confuse short vowel diacritics on  the suffix /ta/, /ti/ /tu/ suffix تاء التأنيث when it refers to first, second, or third person, masculine or feminine, imperative and past tense (كَتَبَ كُتِب كُتُب كَتّبَ كُتّب كتبتَ كتبتُ كتَبتِ كُتِبَتْ كُتّبت). This affects comprehension in L2 learners and causes cacophony and distortion for native speakers and non-native speakers of Arabic. The article sheds light on how AI reads Arabic aloud, classification of pronunciation errors in AI narration; variations in the type and frequency of pronunciation errors across videos; why AI makes mistakes in pronouncing Arabic but does not make grammatical or syntactic errors in AI-narrated content; and how AI pulls off realistic intonation. Additionally, the article gives suggestions for improvement and recommendation for students learning Arabic as a foreign language.