# David AI

> David AI is a San Francisco audio data research company that creates and licenses speech datasets and evaluations for teams building voice AI. Its catalog includes channel-separated conversations, multilingual recordings, group speech and expert dialogue.

- **Founded:** 2024
- **Headquarters:** San Francisco, United States
- **Founders:** Tomer Cohen (Co-founder and CEO), Ben Wiley (Co-founder and CTO)
- **Products:** Converse, Atlas, Chorus, Dialog, Custom dataset development
- **Notable:** Built a library of roughly 100,000 hours of audio in more than 15 languages by May 2025, according to a YC founder interview., Reported more than $10 million in annual revenue run rate by May 2025., Raised $80 million across publicly announced Seed, Series A and Series B rounds.

## Products & services

- **Converse** — Channel-separated, natural English conversations between two speakers across many topics.
- **Atlas** — Multilingual conversation data spanning more than 15 languages, with dialect and accent metadata.
- **Chorus** — Conversations among at least three people for speaker separation and diarization work.
- **Dialog** — Expert conversations across domains.
- **Custom dataset development** — Research-led design, collection, quality review and scaling of audio datasets for customer use cases.
- **Speech-to-speech evaluation** — Human and automated assessments of voice model performance, including a public single-turn preference leaderboard.

## Achievements

- Built a library of roughly 100,000 hours of audio in more than 15 languages by May 2025, according to a YC founder interview.
- Reported more than $10 million in annual revenue run rate by May 2025.
- Raised $80 million across publicly announced Seed, Series A and Series B rounds.
- Published a 2026 speech-to-speech leaderboard using approximately 153,000 comparative ratings across seven model families and 819 prompts.

## Latest updates

- **2026-09** — Released DAI-S2S-ST, a human-preference speech-to-speech leaderboard with 819 prompts and about 153,000 comparative ratings.
- **2026-07** — Published research finding that audio-model judges aligned better with human ratings of content than of acoustic qualities such as naturalness.
- **2025-10** — Announced a $50 million Series B led by Meritech with NVIDIA participation.

## Links

- Website: https://www.withdavid.ai/
- LinkedIn: https://www.linkedin.com/company/david-ai
- Twitter/X: https://x.com/withdavidai

---

Profile page: https://yespress.io/david-ai
Published by YesPress — https://yespress.io
Last updated: 2026-09-29