Datasetpublic

SarahUssama/sada-arabic-test-dataset-sample

🗣️ Arabic Dialect Segmented Speech Dataset (SADA2022 Subset) This dataset contains segmented Arabic speech samples from the SADA2022 corpus, annotated by dialect, gender, age group, speaking rate, environmental condition, and includes ground truth transcriptions. It is intended to support research and applications in Arabic dialect classification, automatic speech recognition (ASR), and spoken language understanding. 📁 Dataset Structure Audio segments are… See the full description on the dataset page: https://huggingface.co/datasets/SarahUssama/sada-arabic-test-dataset-sample.

sourceHugging Faceupdated 1y agoView on Hugging Face

0likes41downloads

Dataset Card

🗣️ Arabic Dialect Segmented Speech Dataset (SADA2022 Subset)

This dataset contains segmented Arabic speech samples from the SADA2022 corpus, annotated by dialect, gender, age group, speaking rate, environmental condition, and includes ground truth transcriptions.

It is intended to support research and applications in Arabic dialect classification, automatic speech recognition (ASR), and spoken language understanding.

📁 Dataset Structure

—Audio segments are stored in .wav format
—Accompanied by a CSV file (dataset.csv) with rich metadata
—Organized into dialect folders:
—Egyptian/
—Saudi/

📊Dataset Statistics

Metric	Value
Total Segments	496
Languages	Arabic (Egyptian, Saudi dialects)
Audio Format	.wav
Sampling Rate	16kHz

📊 Data Insights

🔢 Total Segments: `496`

👤 Speaker Demographics

Attribute	Category	Count
Age Group	Adult (بالغ)	350
	Elderly (كبير في السن)	100
	Young Adult (مراهق)	46
Gender	Male	397
	Female	99
Dialect	Saudi	300
	Egyptian	196

🌍 Recording Conditions

Environment	Count
Clean (نظيف)	217
Noisy (ضوضاء)	154
Music (موسيقى)	125

🕐 Audio Properties

Attribute	Category	Count/Value
Length Type	Short	490
	Long	6
Speaking Rate	Average	320
	Fast	166
	Slow	10
Segment Length (seconds)	Min	0.5
	Max	27.51
	Mean	3.31

📌 Ground Truth Transcriptions

Each audio segment is paired with an Arabic text transcription (GroundTruthText column). There are 493 unique phrases, including:

—السلام عليكم — 3 occurrences
—إلحق يا عمدة — 2 occurrences
—مع السلامة — 2 occurrences
—As well as longer, spontaneous utterances like:

"ايه ولا يهمك ابشر إن شاء الله، ايه الليلة في المزرعة إن شاء الله، نشوفك على خير في أمان الله مع السلامة"

🛠️ How to Use

You can load the dataset using:

python

from datasets import load_dataset

ds = load_dataset("SarahUssama/sada-arabic-test-dataset-sample")