How we test
Last updated
In plain words
- Facts about other products are read from their own pages, with a link and a date. We have not run other products, and the pages say "Not tested by us".
- Results for our own tools come from real runs on this site, with the file, the settings, the browser and the machine named.
- Where no test has been run, we say so. We do not publish estimated results.
Facts about other products#
Each fact about another product on the comparison page and the alternatives pages is read from that company's own page: the pricing page, the help centre or the terms. The fact carries a link to the source and the date we read it.
- A fact older than 180 days is marked as needing a new check.
- A fact older than 365 days is no longer shown.
- If a price sits behind a billing switch or loads in a way we could not read, we leave it out instead of guessing.
- We compare prices only on the same basis: the same currency and a plan billed monthly against our monthly price.
We have not run files through other products. Where a test result for another product would go, the page says "Not tested by us". There is no overall score.
Results for our own tools#
The measured results on the alternatives pages and the comparison page come from runs of our tools on this site, done by us. For each run we record:
- the test file: what it holds, its size in bytes and the start and end of its SHA-256 hash;
- the settings, which are the defaults unless the row says otherwise;
- the size of the input and of the output in bytes, and what the output holds;
- the time from pressing the button to the download, when we measured it reliably;
- the date, the browser and its version, and the machine.
The first runs were made on 16 November 2026 in Chromium 158 on a MacBook Pro with an Apple M3 Pro chip, while the machine was busy with other work. The times are therefore slower than on an idle computer. They are not a speed benchmark. File sizes do not depend on how busy the machine is.
These runs show what our tools did with those files. They do not show that our tools are better or worse than another product, because the other product was not run.
What we have not tested#
We have no measured results yet for caption accuracy by language, text-to-speech voice quality, translation quality or tool speed across devices. The blog articles on those subjects describe a method only. They are kept out of search engines until they hold results from a real run.
Limits of these tests#
A result holds for one file, one browser version and one machine. Your file and device will give different numbers. Each tool page is the place to try your own file.
Corrections#
If a fact here or on a comparison page is wrong or out of date, write to support@neattrim.com. The editorial policy says how we handle corrections.
Questions about this page: see the contact page.