Egyptian Arabic telecom call-center speech data
Telecom customer-service calls the way Egyptian customers make them: bundles, bills, roaming, SIM swaps and network complaints, in Cairene, Alexandrian and Upper Egyptian (Saʿidi) Arabic. Every call is transcribed word for word, tagged with what the caller wants, and checked by several native Egyptian reviewers.
Specification
- Dialect
- Cairene, Alexandrian and Upper Egyptian (Saʿidi) Arabic, with the English words customers mix in
- Calls
- Customer and agent, two speakers per call: bundles, bills, roaming, SIM swaps and network complaints
- Audio
- 8 kHz telephone or 16 kHz wideband, 16-bit WAV, mono or one channel per speaker
- Speakers
- Native Egyptian speakers, balanced across gender and age groups
- Annotation
- Verbatim transcript, timestamps, speaker turns and gender; intents such as activate bundle, bill dispute, roaming, network fault
- Quality check
- Every transcript is cross-checked by several native Egyptian reviewers
Example of the data
- Agent00:01
أهلًا بحضرتك، معاك منى من خدمة العملاء، أقدر أساعدك إزاي؟
Hello, this is Mona from customer service. How can I help you?
- Customer00:04
أهلًا، عندي مشكلة في الباقة، بقالي يومين بحاول ومش نافع.
Hi, I have a problem with my bundle. I’ve been trying for two days and it isn’t working.
- Agent00:09
تمام، ممكن الرقم القومي لو سمحت؟
OK, may I have your national ID number, please?
- Customer00:12
حاضر، بس عايز أعرف ليه حصل كده؟
Sure, but I want to know why this happened.
What it is used for
Speech recognition for Egyptian calls
Train and evaluate ASR that hears Egyptian Arabic, not only Modern Standard Arabic.
Telecom self-service by voice
Let callers change bundles, check bills and switch on roaming without waiting for an agent.
Churn and complaint analytics
Find the calls that end in a cancellation threat, and why.
Agent quality monitoring
Score calls automatically against your service standards.
Questions
Which varieties of the dialect are included?
Cairene, Alexandrian and Upper Egyptian (Saʿidi) Arabic. Tell us the regions and the mix your product needs.
Which telecom intents are labelled?
Intents such as activate bundle, bill dispute, roaming, network fault, plus sentiment and named entities on request, or your own label set.
Can I hear samples first?
Yes. Request samples with the form on this page and we will send them with the full specification.
Can it be used in a commercial product?
Yes. The data is prepared for training and evaluating commercial AI systems, under a licence agreed with you.
Tell us what your model needs.
Request samples and the full specification. We reply within a couple of business days.
Request samples