Export a transcript

File ▸ Export Transcript… listens to your tracks and writes what was said to a file. It happens on your Mac: your audio is not sent anywhere for it.

It needs macOS 26 or later. On an older Mac the item is still in the File menu, dimmed, and reads Export Transcript… (Requires macOS 26).

Like exporting audio, it is part of the free trial and needs a subscription after it.

How it works

Each track is treated as one speaker, so PodEditor does not have to guess who is talking: the words on the Brandon track are Brandon’s. That is why a project with a separate track for each person works best. A project with everyone on one track still works, but the transcript has no speaker names.

The transcript follows your edit. Text you cut is not in it, and a clip you moved takes its words along. Muted tracks start unchecked, and muted clips are skipped. A track of sound effects or music starts checked like any other, because PodEditor cannot tell what a track holds; untick it, or it is transcribed too and may produce stray words. The dialog remembers what you ticked for next time.

The dialog

File ▸ Export Transcript… opens one save panel, with the options in it:

• Tracks. Tick the tracks to transcribe. Each has a speaker name, filled in from the track’s name. Change it here if the track name is not what you want in the file; this does not rename the track or change your project. Every selected track needs its own name, and a name cannot contain a colon or a line break.
• Language. Pick the language of the speech. PodEditor cannot detect it, and a wrong language gives text that looks plausible and is nonsense, so check the sample on the last screen.
• Format. Text (the name on its own line, then [1:20] words), SubRip (.srt) or WebVTT (.vtt). Changing the format changes the file’s extension in the name field. In a Text file each speaker’s continuous speech is one paragraph, so a “Mm-hmm” in the middle of someone’s sentence is its own short paragraph after it, not a break in it. A long stretch of one person talking is cut at the end of a sentence after about 110 words. The caption files follow the Podcast Index recommendations: about 32 characters by two lines per card for SRT, 65 characters per line for VTT with the speaker in a voice tag. They start a new card at the same places the Text file starts a new paragraph, so the three files group the speech the same way. The transcript matches the audio: “um” and “uh” stay in, because PodEditor does not cut them out of the sound.
• Remove Mic Bleed is for tracks recorded in the same room, where each microphone also hears the others. It is off by default because on separately recorded tracks it removes quiet real speech such as “Mm-hmm”. Hover over it for the short version.
• Reset Transcription… deletes the transcription kept in the project (not your exported files). To transcribe again from scratch, reset it and export.

The panel opens in the folder that holds the project. If the names or language cannot be used, Export says why and the panel stays open.

The first time you use a language, your Mac downloads its speech model from Apple. The panel says This language needs a speech-model download first. (nothing is shown when the model is already installed), and the transcription shows “Downloading the speech model…” while it happens. It needs an internet connection once; after that it works offline.

While it runs

A sheet shows progress. You can use other apps while it runs; PodEditor itself is locked, so you cannot edit, save, open another project or export until it finishes or you press Cancel. Quitting or closing the project asks first and stops it. A 74-minute two-person episode took about a minute and a half on an M4 Mac mini; an older Mac takes longer.

Check it before you publish it

It is a draft. Proper names and jargon are the usual misses (“Kubernetes” has come out as “Cuber Needs”). The last screen shows how each speaker’s words begin; read them against what was said. When people talk over each other, the overlap shows up as short attributed fragments, and the caption files show the overlapping cards.

Saved recognition

The recognition is saved inside the project, in a transcripts folder, so the next transcript, in another format or with different names, takes a moment instead of a minute. It covers each track’s whole audio, including parts you cut, so a copy of the project carries that text too. Use Reset Transcription… in the dialog if you do not want it there. Your remembered choices (names, language, format) are kept separately and are not cleared with it.