Speaking Diary

Privacy

Updated November 2026

This is an early version of our privacy notice. It will be reviewed before public launch.

Your voice and words stay on your device

The Speaking Diary listener is an AI model that runs inside your browser. When you talk, your voice is processed on your computer. We don't receive your voice, your words or the listener's replies, and we can't read them.

Clips

After a call you can make a short video clip. It is made on your device: your words are written down by the same on-device model, and your voice is in the clip only if you choose to include it. The clip leaves your device only if you share or download it yourself, and the call's audio is forgotten when you leave the page.

What gets downloaded

When you first open any page on a computer that can run it, your browser starts downloading the AI model (about 2.5 GB) in the background from Hugging Face (huggingface.co) and stores it on your device, so later visits start quickly. Phones don't download it. Like any website, Hugging Face and our hosting provider (Amazon Web Services) receive standard request information such as your IP address when your browser fetches files.

Your account

Your first 3 calls need no account. After that you sign in with Google or Facebook, with no password. We then keep:

We never keep your voice, your words, the listener's replies or how you rated your mood. These are stored with Supabase, our database provider, on Amazon Web Services in the United States. We use them to run your account and to count calls, and we don't sell them or use them for advertising.

Deleting your account

On the talk page, open the account menu (your initial, top right) and choose Delete account. This deletes your name, email and call history straight away. If you signed in with Facebook, removing Speaking Diary in your Facebook settings only stops future sign-ins; use Delete account to delete your data.

What stays on your device