Description
We’re running a short video data collection project for one of our clients.
We’re looking for native Arabic speakers across multiple Arabic-speaking countries and dialects to record a 15–20 second video, speaking naturally in Arabic and looking directly into the camera.
The videos collected will be used to help train and evaluate an AI model that distinguishes real, human-recorded video from AI-generated (synthetic) video.
Project Details:
- Start Date: ASAP
- Duration: 1 session
- Commitment: 1 hour.
- Job Type: Freelance
- Location: Remote
- Language/Dialects: Yemeni Arabic
- Pay Rate: $4.00-$15.00 (Depending on country location)
Eligibility
- 18 years of age or older.
- Native Arabic speaker (self-identified).
- You may be asked to complete a brief screening step to confirm your dialect before your submission is finalized.
What you’ll do Record a single 15–20 second video in your native Arabic dialect. Speak naturally (no script). Suggested prompts: briefly introduce yourself, mention where you live, or talk about an interest/hobby. Look directly at the camera the whole time. After recording, answer a few short questions about yourself (e.g. age range, country, dialect) to help us organize the dataset.
Recording specifications (must meet all to be accepted)
- Video quality: 1080p or higher
- Length: 15–20 seconds
- Framing: Head-and-shoulders, face centered and fully visible
- Lighting/background: Well-lit face (no harsh shadows), plain/neutral background
- Expression: Neutral/relaxed while speaking naturally
- Audio: Clear speech, no background noise, music, or echo
- Orientation: Portrait or landscape acceptable (hold steady, no shaky footage)
- Environment: Quiet indoor setting preferred
- Prohibited: Filters/beauty modes, sunglasses, hats/hoods, face coverings, headphones/earbuds, watermarks, stickers, borders, other people/pets in frame
Your Privacy Before you begin, you’ll be asked to review and accept a consent form explaining how your video and any information you provide will be used, stored, and who it will be shared with.