Published by PitchChanger · Updated · How these guides are made
How to use this workflow
Open storage before choosing a file
Use a harmless, distinctly named local recording. Open developer tools and inspect this origin under Application or Storage. Record Cache Storage and IndexedDB entries before loading it. Do not clear anything yet: that would erase the baseline.
Compare after one operation
Import the harmless file, run one tool and inspect the same storage views again. Record changed keys and sizes. A new entry is a lead to inspect, not proof that it contains source audio.
A storage classification worksheet
An original proposed inventory separates engine assets, model weights and personal content. A hypothetical cache key ending in model.onnx suggests model weights; its name alone does not prove the payload. Inspect the request URL, response type and application code where available. A WAV-looking key may be a public demonstration rather than your input.
Swipe or scroll sideways to read all columns.
| Entry observed | Evidence to collect | Conclusion allowed |
|---|---|---|
| Model URL | Origin, bytes, response type | Likely reusable model; verify contents |
| Named input key | Before/after change and payload | Investigate possible persistent input |
| No changed keys | Storage types actually inspected | No change observed in this scope |
Verify the result and record its limits
Repeat after closing and reopening the tab. Memory-only objects disappear with the page, while persistent entries may remain. Absence from Cache Storage does not rule out IndexedDB, another origin, extensions or files on disk. Report which stores you inspected rather than declaring that nothing was retained.
Common questions
Was this example measured on my device?
No. The worksheet and any hypothetical numbers are explicitly labeled. Use a harmless recording to collect your own observations and record the browser, date and exact operation.