Privacy

Four things, and nothing else:

• The email address you sign in with, and a Google account id if you sign in that way.
• Your briefs and pack settings: the text you typed, the count, the length, the category, the filename pattern, the loudness and format targets.
• The audio that comes back from a render, as WAV files in private storage.
• One ledger row per render, holding the model, the number of audio seconds and what the provider charged. It is how the allowance is counted.

There is no analytics, no tracking pixel and no third-party script on this site. The only cookie Sona sets is the sign-in session.

In Supabase: Postgres for the rows, Supabase Storage for the audio. The project is hosted in the EU region, in Frankfurt. The site and the API run on Vercel, which keeps short-term request logs containing IP address and path.

The audio bucket is private and no row in the database is readable by any account other than yours.

Replicate, which hosts the generation models, receives the prompt text derived from your brief and the render parameters. It does not receive your email address, your account id or anything else about you.

If you sign in with Google, Google knows you signed in to Sona. If you sign in with a code, the email is sent through Supabase.

Nobody else. Sona does not sell data, does not share it with advertisers, and does not use your briefs or your audio to train anything.

Packs, takes and audio stay until you delete them or delete your account. Ledger rows are kept for as long as the account exists, because they are the only record of what early access cost.

Deleting a pack deletes its takes and their audio. Write to hello@sona.audio to have the whole account removed; the profile, its packs, its audio and its ledger go with it. Ask for a copy first if you want one, and it will be sent as files.

Audio you render is yours to use commercially, including in paid client work and in products you sell. Sona claims no rights to it and requires no credit.