NileTTS: 38 Hours of Egyptian Arabic Text-to-Speech Data

NileTTS provides 38.1 hours of transcribed Egyptian Arabic speech across 9,521 utterances, published by KickItLikeShika in February 2026. The collection is segmented into specific domains, including over 21 hours dedicated to sales and customer service interactions.

Use Cases

• Training Egyptian Arabic TTS models using the 38.1-hour audio corpus
• Building automated customer service agents using the 4,975 utterances in the Sales & Customer Service domain
• Evaluating dialectal speech synthesis using the 950-utterance evaluation set

Strengths

• 38.1 hours of audio
• 9,521 transcribed utterances
• Apache 2.0 license
• Domain-specific labels for sales and general conversation

Limitations

• Contains synthetic data
• Geographic bias toward Egyptian dialect
• Domain imbalance with 52% of utterances focused on sales

Provenance

Source

KickItLikeShika (Arxiv 2602.15675)

Collection Method

Synthetic data generation and transcription

Freshness

Last updated February 2026

Geography

Egypt