Gulf Arabic–English code-switching speech data
How the Gulf really talks at work: Saudi, Emirati, Kuwaiti and Qatari Arabic that switches to English mid-sentence, “خلنا نسوي meeting بكرة”. Every word is transcribed in its own script and every switch is tagged, so recognition stops breaking at the language boundary.
Specification
- Dialects
- Saudi (Najdi, Hejazi), Emirati, Kuwaiti and Qatari Arabic mixed with English
- Settings
- Meetings, sales calls, banking and telecom calls, interviews and everyday conversation
- Audio
- 8 kHz telephone or 16 kHz wideband, 16-bit WAV
- Annotation
- Each word in its own script (Arabic or Latin), language tag per word, switch points, timestamps and speaker turns
- Speakers
- Native Gulf speakers who use English at work, balanced across gender and age groups
- Quality check
- Several bilingual reviewers check every transcript
Example of the data
خلنا نسوي meeting بكرة الصبح ونقفل الـ deal.
Let’s have a meeting tomorrow morning and close the deal.
الـ deadline يوم الخميس، فلازم نخلص الـ report قبل.
The deadline is Thursday, so we have to finish the report before.
What it is used for
Speech recognition that survives switching
Train ASR that keeps both languages right inside one sentence.
Meeting and call transcription
Transcribe Gulf meetings and sales calls without losing the English terms.
Voice agents for business users
Understand customers who say “transfer” and “statement” in English inside Arabic sentences.
Benchmarks
Measure how well your model handles switching, word by word.
Questions
How is mixed speech written down?
Each word in its own script: Arabic words in Arabic letters, English words in Latin letters, each tagged with its language.
Which dialects switch to English here?
Saudi (Najdi and Hejazi), Emirati, Kuwaiti and Qatari Arabic. Tell us the mix your users speak.
Can I see samples first?
Yes. Request samples with the form on this page and we will send them with the full specification.
Related datasets
Tell us what your model needs.
Request samples and the full specification. We reply within a couple of business days.
Request samples