Speaker Label in Transcription

The audio is analysed for changes in voice rather than for names. When one voice stops and another starts, a boundary is placed and a new label is assigned. Nothing is known about the people themselves; the labels are numbers that the transcript keeps consistent from the first turn to the last.

Yes. Each label can be renamed once, and every turn carrying that label changes with it. A recording that starts with Speaker 1 and Speaker 2 can end with the two names used in the meeting, so the final transcript reads as a conversation rather than a numbered log.

Overlapping speech is the hardest case. One turn usually wins, and the other may be shortened or attached to the wrong label. A quick read of the affected passage is worth the time, since a single mistaken attribution changes who is credited with a remark. Cleaner audio reduces how often this happens.

Panel recordings with several participants are handled, and each distinct voice receives its own label. Very large groups recorded on a single microphone blur together, so a session with more than a handful of people benefits from a transcript review. Two-person interviews are the simplest case.

Yes. The speaker tag travels with the text into the exported file, whether that is a plain document or a subtitle file. Renaming happens before export, so the file that leaves the editor already carries the names a reader expects instead of placeholder numbers.

No separate charge applies. Speaker separation is an option in the transcription settings, switched on before the job starts. Leaving it off produces a continuous transcript without labels, which is often enough for a single narrated recording and slightly faster to read.

Yes, and it usually should. A lecture, a voice memo, or a solo presentation gains nothing from numbered turns. With separation switched off, the transcript reads as paragraphs, which is easier to skim and to paste into a document than a list of labelled one-line turns.

Distance from the microphone matters more than the file format. Participants recorded on one room microphone at uneven volume produce fewer reliable boundaries than a call where each person has a separate input. Background music and room echo also blur the switch between voices.