ActiveREQ-000006
Unscripted Two-Person Conversational Audio-Video Dataset
Synchronized, consented recordings of natural conversations between two adults.
- Data type
- Audio and Video
- Volume
- Not specified
- Delivery date
- Not specified
Requirement brief
Specific requirements
- Two adults engaged in a natural, friendly conversation as friends or family members, not strangers
- Webcam recording at 1080p or higher and 30 fps, with faces clearly visible and good lighting
- 48 kHz audio with one microphone per speaker and separate audio tracks
- Four files per session: Speaker 1 video, Speaker 2 video, Speaker 1 audio and Speaker 2 audio
- Audio and video synchronized within 40 ms
- Session duration: 10–60 minutes
- Multiple languages with a balanced distribution
- Verbatim transcripts with speaker labels and timestamps
- Signed participant consent permitting commercial and AI/ML use
- Participants must be 18 years or older
- Maximum contribution of three hours per participant
- No scripted conversations, AI-generated content or minors
Coverage
Required coverage
- Multiple languages
Interested in this requirement?
Review the supplier process and prepare your capabilities.