Sound familiar?Getting speaker attribution wrong in multi-person scenes?
Multi-Speaker Conversation (Audio + Video)
Real broadcast scenes where 2–6 people talk over each other. Everyone wore their own mic, so each voice is on a separate track — and each voice is linked to the face that spoke it.
Use cases
Problems this data solves
Fix speaker attribution in multi-person scenes
Train lip-sync on overlapping dialogue
Generate dialogue in Korean and other low-resource languages
Scale · specs
- Total volume
- [Hours] · [Clips]
- Speakers
- 3–[N] people, Korean
- Genre
- Variety · talk · [other genres]
- Audio
- Per-speaker WAV, [sample rate], original mic / AI-separated flagged
- Video
- Original-resolution MP4, [fps]
Labels
- Included
- Transcript, speaker timeline, 3-level captions, speaker metadata
- On request
- Emotion, overlap markers, other custom labels
Request a sample
Get a sample of this collection
The button takes you to the sample request form.
Delivery
How you receive it
File formatMP4 + per-speaker WAV + JSON
Bundle unitPer scene (clip)
Delivery methodDownload link · cloud bucket (manifest included)
Rights & consent
Rights & consent
Direct contract with the rights holder
Contracted directly with the producer who owns the originals
Performer consent scope
Within the scope of performers’ consent for AI training
Face handling options
Blur non-consenting faces or fully anonymize
Non-exclusive · fixed term
Non-exclusive use within the contract term
No-replication clause
No replicating performers’ faces or voices
Need something slightly different?
Change the speaker count, genre or labels and request a custom build.