Never leaves
The audio bytes. The unreleased master. Anything a model could train on as music. Thermal cannot store a copy it does not receive.
Proof
The file never uploads. We cannot train on your music because we never get a copy. That is an architecture, not a policy.
The airplane-mode check
Open thermal-ai.app, then turn on airplane mode, or switch wifi and cellular off.
Drop the file on the homepage instrument. Tempo, key, and the rest of the read print from this page, in this browser.
If the page was already loaded, the worker is already here. The decode never waits on a server.
If the tab was not cached and you are already offline, reload after you are back on the network, then repeat the check. The measurement itself does not need the network.
The network log
| Request | Carries the audio | Needed for the read |
|---|---|---|
| None to Thermal | No | No. Decode and measurement stay in the page. |
| Analytics ping, if you are online | No | No. Path and event name only. Not the file, not the fingerprint. |
| Ad tag, if you are online | No | No. Marketing pages load it. The worker does not. |
The honest split
The audio bytes. The unreleased master. Anything a model could train on as music. Thermal cannot store a copy it does not receive.
A compact measurement object, if you start a written kit, a mix read, sonic neighbors, or a public page: tempo, key, loudness as dBFS, word labels, and a short energy envelope. Not the file.
The distinctiveness score stays in the studio. It is not printed on a public page, and it is not how the directory is ordered.