I sometimes have an audio file that is already finished — an interview, voice memo, lecture, or exported call. I do not want to play it into Dictation or upload it to a transcription website.

I put the audio file in its own folder, import that folder into MinuteFile for Mac, and transcribe it there. MinuteFile 0.1.8 runs the transcription on the Mac and keeps the recording beside an editable transcript in its local archive.

Prepare one completed audio file folder

This workflow starts after the recording exists. It is not live capture, microphone dictation, or a meeting bot.

If the source is a completed Zoom recording, use the separate Zoom recording workflow to find the local or downloaded meeting files first. The steps below are for an audio file you already have on the Mac.

MinuteFile recognises M4A, MP3, WAV, AAC, FLAC, and OGG audio files. Before importing one, play a short section in QuickTime Player or another local player. Check that the file opens, contains the speech you expect, and has enough volume to hear the speakers.

Clear audio matters. Local transcription cannot recover words hidden by clipping, loud music, or several people talking at once. Keep the original file even if you make a cleaner working copy because you may need to compare the transcript with the source later.

MinuteFile imports recording folders, not loose files. Create a folder in Finder and give it a useful name, such as:

Interview with Robin - 2026-08-01

Move or copy the completed audio file into that folder. One audio file is enough. If you also have captions, a transcript, or notes for the same recording, keep those files in the folder too.

I use one folder per recording because the folder becomes the record in the archive. The name is easier to search later than audio-final-2.m4a, and supporting files stay attached to the right source.

Import the recording folder into MinuteFile

Open MinuteFile and click Import Recording Folder. Choose the folder you just made, not the audio file inside it.

MinuteFile adds the recording to its inbox. Select it and confirm that the audio file appears in the recording details. If MinuteFile reports that it cannot find a supported recording, check that you selected the containing folder and that the audio file uses one of the supported formats above.

Importing the folder does not start live recording. MinuteFile only works with the completed files you chose.

If MinuteFile does not show the audio file

Check the folder in Finder before changing any transcription settings. The audio file should have finished copying, appear directly inside the folder you selected, and play normally in another Mac app.

Then check the filename extension. Renaming an unsupported file to end in .m4a or .wav does not convert its audio data. Export a supported copy from the app that created the recording, keep the original, and import the folder again.

If the folder contains several unrelated recordings, separate them first. One folder with one completed recording gives the archive a useful name and makes it clear which source file belongs to the transcript.

Choose the language and transcribe on the Mac

Select the spoken language and transcription model shown in the transcript toolbar, then click Transcribe. If the chosen model is not installed, MinuteFile offers to download it first.

That model download is a network request. Once the required model is installed, MinuteFile sends the audio to its local transcription process. The recording and transcript are not uploaded to MotionObj or a cloud transcription service.

Leave the source file in place while transcription runs. When MinuteFile finishes, it opens the saved transcript with timestamps connected to local playback.

The transcript is saved in MinuteFile’s local archive. You can quit the app, reopen the recording later, and continue reviewing without retranscribing the audio. The original audio file still belongs in your own backup rather than relying on the app archive as its only copy.

Review the transcript while listening to the file

I do not treat automatic transcription as final copy. I play the local audio from the transcript workspace and correct names, technical terms, punctuation, and any words the model missed.

Clicking a known timestamp moves local playback to that part of the recording. Add a bookmark when a passage needs another listen or when you want to return to it later.

Transcript and speaker labels can be wrong, especially with background noise or overlapping voices. Review anything you plan to publish, quote, or keep as an official record.

Assign People manually when names matter

MinuteFile does not identify voices or infer who a speaker is. You can edit a meeting-only speaker label or create a reusable Person and assign that Person yourself.

I keep a meeting-only label when I do not know the speaker. I assign a Person only after I recognise the voice from the recording or have another reliable source for the name.

If one transcript turn contains two speakers, place the insertion point where the speaker changes and use Split Turn. MinuteFile preserves the original known timestamp on the first block and marks the derived block as untimed. It does not invent a timestamp for the split or guess the new speaker.

Search the archive and export a copy

The saved transcript remains editable and searchable in MinuteFile. Archive search is literal, so search for the exact word or phrase you expect. You can also filter results by a Person you assigned manually.

Export a TXT, Markdown, or HTML copy when you need the transcript outside the app. I keep the source recording and exported copies in my normal backup system; installing or removing the app does not replace a backup.

Local processing does not change consent, confidentiality, or copyright rules. Make sure you are allowed to keep and transcribe the audio file.

Use an existing transcript instead of transcribing again

If the folder already contains VTT, SRT, timestamped text, or plain TXT, MinuteFile can use it as the editable archive transcript without running transcription again.

Recognised VTT, SRT, and timestamped-text timestamps stay connected to local playback. Plain TXT remains editable and searchable, but MinuteFile does not pretend that plain text contains timing information. The original imported transcript remains labelled as a Supporting File.

What stays local and what still uses the network

Recordings, transcripts, People, assignments, and archive search stay on the Mac. MinuteFile does not upload them to a MotionObj backend or cloud transcription API.

MinuteFile still uses the network for optional transcription and speaker-model downloads, signed update checks and downloads, and Stripe checkout when you choose to buy a licence. The MinuteFile privacy policy documents those boundaries.