What it sells
Training Data For Speech Models.
Business model breakdown
Buyer: who owns the budget?
UnverifiedData/research leads at frontier labs and teams training speech and voice-agent models.
Trigger: what makes them buy now?
Speech models trained on audiobooks, podcasts and synthetic dialogue fail at intent and turn-taking; need real conversational audio.
Workflow: what inputs become what output?
Records real people in scripted or free conversations, delivers per-speaker 48 kHz stems with word-timed transcripts and event logs (tool calls); catalog includes tool-calling, accented English, multilingual, multi-party, turn-taking evals.
Economics: what it saves and how it charges
Sells datasets under commercial license; one free public dataset (Open Yap 1K) as lead-gen.
Distribution: how it reaches customers
"Request access" per dataset and contact form; private datasets for buyers; public dataset on Hugging Face/GitHub.
Funding history and coverage
| Round date | Stage | Amount | Investors | Source |
|---|---|---|---|---|
| — | Y Combinator Fall 2026 | — | Y Combinator | ycombinator.com |
Profiles summarize public reporting and company websites. Business-model notes are MaxBC's reading of that material, not confirmed facts; unknowns are left blank rather than guessed. Updated 2026-10-12.