Layla Witheeb: Jordanian Arabic Acoustic Dataset

This dataset presents comprehensive phonetic data collected from 109 male and female undergraduate students representing most governorates in Jordan. The primary objective of this repository is to provide robust empirical acoustic data on Jordanian Arabic in two different speaking conditions: story-reading and storytelling. The research hypothesis driving this data collection is that the temporal and spectral properties of speech segments may be altered due to higher temporal and communicative demands in the storytelling condition. Furthermore, this dataset aims to facilitate the study of Jordanian Arabic dialects across the various regions represented.

Data was collected through a reading task of the story "Little Red Riding Hood" (Layla Witheeb) by the 109 native speakers of Jordanian Arabic. The speakers were subsequently asked to tell the story from memory (storytelling task). Both the read and told texts were then transcribed in Arabic orthography and then into phonetic ATR transcription. Audio recordings were forced-aligned and segmented automatically using WebMaus, and then manually verified and corrected. Acoustic measurements can be extracted using the Praat script provided with this database.

The repository contains the raw data organized according to the speakers' geographic regions. Each regional folder contains individual speaker folders housing the respective audio and text files. A master metadata file is provided in the root directory, detailing all relevant information about the speakers. Researchers in linguistics, phonetics, and Arabic dialectology can utilize this dataset to conduct primary or secondary analyses, use it as training data, or perform cross-dialectal and cross-linguistic comparisons.