Articles on this site are signed by the team rather than by individuals. The people behind the byline build and support GPT Transcribe day to day, which is where the material comes from — support threads, real recordings that failed in interesting ways, and the settings that turned out to matter. Treat it as an organizational byline: not one named author, and not an independent testing lab.
Two areas. First, the practical path from a raw file to text you can hand off: what to do about a distant microphone, when speaker labels are worth enabling, which export format survives the next step. Second, plain-language coverage of the speech and audio models people ask us about — what each is built for, where the provider's claims end and our own observation begins, and dates for both.
The editorial policy spells out what we verify before publishing, how vendor figures are attributed, and when a page gets revised. Spotted something wrong, or want a claim sourced? Send the URL to support@gpt-transcribe.org and it goes to the same people who wrote the page.