Kratos-AI/Egyptian_People_Speaking_Video_Dataset

Egyptian People Speaking Video Dataset

This dataset contains high-quality video recordings of Egyptian people speaking on a range of topics. It is curated for AI research in speech recognition, multimodal analysis, topic understanding, and spoken-language modeling.

Contact

For queries or collaborations related to this dataset, contact:

• [email protected]
• [email protected]

Supported Tasks

• Task Categories:

  • Video Classification
  • Speech-to-Text (Automatic Speech Recognition)
  • Multimodal Understanding

• Supported Tasks:

  • Transcription of spoken Egyptian Arabic or bilingual (Arabic-English) speech
  • Topic classification based on speech content
  • Sentiment and emotion analysis from spoken video
  • Speaker diarization and identity-agnostic speaker recognition
  • Multimodal analysis combining audio, video, and gestures
  • Research in human-computer interaction and AI communication models

Languages

• Primary Language: Egyptian Arabic
• Secondary Presence: English (if code-switching occurs)

Dataset Creation

Curation Rationale

The dataset was created to support AI systems in understanding natural spoken Egyptian Arabic, enabling speech-to-text, topic modeling, and multimodal interaction research. It is intended for educational, accessibility, and research purposes.

Source Data

• Contributors: Volunteers and participants recorded speaking on assigned topics
• Collection Process: Videos were recorded in controlled or natural environments. Personal identifiers unrelated to the speech task were removed or anonymized. All participants gave explicit consent for research use.

Other Known Limitations

• Bias: Urban dialects and younger speakers may be overrepresented
• Environmental Noise: Background noise may affect speech clarity
• Topic Coverage: Selected topics may not fully reflect all cultural or regional contexts in Egypt

Intended Uses

✅ Direct Use

• Training Automatic Speech Recognition (ASR) systems for Egyptian Arabic
• Topic classification and sentiment analysis research
• Multimodal AI studies combining video, audio, and gestures
• Human-computer interaction and accessibility applications

❌ Out-of-Scope Use

• Surveillance or identification of individuals without consent
• Commercial exploitation without ethical clearance
• Any use violating participant privacy or recording agreements

License

CC BY 4.0