Sayso Voices ยท Consent

What you agree to.

Sayso Voices collects recordings of people who read and say short texts. The recordings train open speech and text models for Sayso, a dictation app. This page says what we collect, how we use it, and what you can do.

Who can take part

You must be 18 or older. You record only your own voice.

What we collect

We do not ask for your name, your email address, or an account.

How we use it

Your choice decides the use. With Train Sayso models, your recordings train and test Sayso models, and nobody else gets the audio. With Train Sayso models and publish, your recordings and their text also go into the public Sayso dataset under the license CC BY 4.0, so anyone can use them, with credit to Sayso.

The Sayso models are open. Anyone can download them. A model learns from the recordings, but it does not contain the recordings.

Where it is kept

The recordings and their data are kept by Cloudflare (R2 storage and the D1 database) for Watzon Ventures LLC, the company that makes Sayso. While you speak, the audio also goes to a Deepgram speech model that Cloudflare runs (Workers AI), to show the words live. That model keeps no data, and we opt out of the Deepgram Model Improvement Program.

Your rights

Your key is the only proof that the recordings are yours. If you lose it, we cannot find your recordings for you.

Version 2026-11-05-draft.

Start reading