For AI teams
Dataset specs.
Verified entries show dated counts from FIUND’s audited catalog. Sourced-to-brief pages describe collections we can commission but are kept out of search until real inventory exists. Every delivery still passes buyer-specific rights, consent, and privacy review.
The master collection and its speaker subsets overlap. Subset figures are included in the master total and are not additive.
English multimodal podcast conversations
A cataloged collection of 1,058 long-form English recording sessions with aligned audio, video, and transcript assets. T…
English two-speaker podcast conversations
A 867-session, 192.72-hour two-speaker subset of FIUND’s cataloged English multimodal podcast collection, with audio, vi…
English single-speaker podcast speech
A 174-session, 32.78-hour single-speaker subset of FIUND’s cataloged English multimodal podcast collection, with audio, …
Conversational Brazilian Portuguese speech (sourced to brief)
Natural, multi-speaker Brazilian Portuguese conversation recorded in real conditions, sourced directly from owners with …
US English contact-centre conversations (sourced to brief)
Task-oriented US English phone conversations with consent and PII handling in place, sourced to a buyer’s spec for voice…
Egocentric kitchen video for robotics (sourced to brief)
First-person video of everyday kitchen tasks captured with wearable cameras and consent, for robotics manipulation and w…
Consented motion-capture library (sourced to brief)
Rig-ready motion capture from consented performers, licensed for avatar animation, gesture synthesis, and pose work. Per…