Datasets
Six categories we collect every week. None are off-the-shelf — every project is scoped, recorded, and QA'd to your requirements.
Natural, overlapping dialogue between two or more speakers — diarized, turn-segmented, and tagged.
Common use cases
Long-form, expressive reads from a single speaker — controlled prompts, consistent capture.
Common use cases
Synchronized multi-mic, multi-camera capture for multimodal research — viseme, gaze, and gesture friendly.
Common use cases
Native speakers across 40+ languages — collected by humans, never machine-translated or synthesized.
Common use cases
Region-tagged accents within each language — from London RP to Lagos English to Chicano Spanish.
Common use cases
Every recording shipped with structured metadata — speaker, environment, device, emotion, and consent.
Common use cases
Custom is the default. Tell us what your model needs and we'll scope it.