DATA WE WORK WITH

The right source.
A clearer signal.

Explore real-world data for training and evaluation. Find a starting point by data type or language, then tell us the task and coverage you need.

Discuss a data brief ↗
Browse capabilities below. For named collections, explore the dataset catalog →

Speech & languages

Conversation, recording context and regional language requirements.

Video & human activity

First-person tasks, recorded knowledge and visual environments.

Physical-world & sensor data

Capture perspectives, movement, depth and aligned sensor streams.

Text, documents & images

Written content, business records and creative source material.

Music & audio

Recordings, production components and spatial audio.

Custom collection

A focused pilot for a specific task, setting or participant group.

CUSTOM COLLECTION

Some tasks need
a new source.

If the right material doesn’t exist yet, start with a focused capture brief. Define the people, setting, format and criteria for a useful pilot.

Explore custom collection ↗
FIUND / CAPTURE BRIEF
01
Define

Task, people, setting

02
Pilot

Capture protocol, examples

03
Review

Acceptance criteria

04
Expand

Agreed collection scope

ILLUSTRATIVE FORMAT Scope agreed per project