FileDial › Guides › Video and subtitles
How to transcribe audio or video to subtitles on a Mac
By Himanshu Ranjan · Updated 8 November 2026
Select an audio or video file in Finder, open FileDial, and run Convert › Transcribe to subtitles… It makes an SRT and a TXT on this Mac, so the recording is never uploaded. Automatic transcripts contain mistakes, so proofread names, numbers and unclear passages before using them.
What you get
FileDial's Convert › Transcribe to subtitles… accepts audio and video and produces two files: an SRT, which has cue timings for use as subtitles, and a TXT with the words. Both are made on this Mac.
That covers interviews, lectures, voice memos and video narration, without sending anything to a transcription website.
Transcribe with FileDial
- Select the audio or video file in Finder.
- Hold the right Option key to open the dial.
- Choose Convert › Transcribe to subtitles… and follow the prompt.
- Open the SRT in a text editor or player, and read the TXT.
Improve the transcript's odds
Speech recognition does better on clear audio. Before transcribing a noisy recording, run Clean › Clean up voice, or Clean › Remove background hum, and use Edit › Normalize loudness if the level is low. Edit › Trim… removes long irrelevant sections. Automatic captions are generally less accurate than human work, especially with homophones, so proofread.
Next steps with the SRT
Fix timing with Edit › Shift timing… Number cues correctly with Edit › Fix numbering. Translate with Translate › Translate subtitles… Attach to a video with Combine › Add subtitle track, or Combine › Burn in subtitles for permanent text. The guide on adding subtitles to video walks through that last step.
Be honest about accuracy
FileDial's note describes where the transcript is made, not how accurate it is, so no accuracy figure is claimed here. Always review text that will be published, quoted or used for accessibility.
What affects transcript quality
Automatic transcription depends heavily on the recording. Clear speech with one speaker close to a microphone is the easy case. Crosstalk, music under the voice, room echo, distant microphones, strong accents, technical terms and names are the hard cases. Wikipedia notes that automatic captions are often less accurate than human work, especially with homophones, and that disability-rights groups want them human-reviewed when they are used for accessibility.
You cannot fix a bad recording in the transcript, but you can often improve the input. Cut out sections you do not need, reduce hum and noise, and bring a low recording up to a reasonable level before transcribing.
A practical workflow
First, make a working copy of the audio. If the file is long, Edit › Trim… can remove the preamble. Run Clean › Clean up voice if there is rumble or noise, and Edit › Normalize loudness if it is quiet. Then run Convert › Transcribe to subtitles…, which makes an SRT and a TXT. Read the TXT first for a quick overall check, then play the SRT with the media and correct errors in a text editor. Fix names and numbers carefully, since errors in those are the most costly.
If the subtitles are for a video, attach them with Combine › Add subtitle track, which needs a video and a subtitle file and does not re-encode, or with Combine › Burn in subtitles for text that is part of the picture. If viewers need another language, translate the corrected SRT, not the raw one.
Mistakes, limits and alternatives
The main mistake is publishing without proofreading. Another is correcting only the TXT while the SRT used for the video still has the old errors, so correct the SRT and make the text from it with Convert › Convert to plain text. A third is expecting speaker labels, which a plain SRT does not carry.
Alternatives on a Mac include dictation features and online transcription services. Dictation is built for live typing rather than transcribing files, and online services upload your recording, which is the trade-off FileDial avoids by working on this Mac. The right choice depends on whether privacy or a human-reviewed service matters more to you.
Proofreading efficiently
Play the audio at normal speed with the SRT open, and pause only where the text looks wrong. Fix proper nouns first, then numbers and dates, then punctuation. A search for common mishearings, once you spot them, fixes repeats in one go. Read the TXT aloud for the final pass, because awkward sentences stand out when spoken. If the speaker has an accent or uses jargon, make a short glossary and check every instance. When the corrections are done, regenerate the plain text from the corrected SRT so both files agree.
Private by design
Recordings of meetings and interviews are exactly what you do not want on a stranger's server. FileDial runs every tool on your Mac and never uploads a file, and all tools work with Wi-Fi off.
Questions
Where does transcription happen?
On your Mac. FileDial's Transcribe to subtitles… makes an SRT and a TXT locally, and files are never uploaded. It works with Wi-Fi off. The recording stays on your drive throughout, and no network connection is needed for the transcription itself.
How accurate is it?
No figure is claimed. Automatic transcription is generally less accurate than human work, especially with unclear audio, names and homophones. Proofread before publishing. Wikipedia notes automatic captions are often less accurate than human work, so assume corrections will be needed.
Can it transcribe a video?
Yes. The tool accepts audio and video files. Select the video in Finder and run Convert › Transcribe to subtitles… Select the video file and run the tool; the sound is used to make the text, and the video itself is not changed.
What do I do with the SRT afterwards?
Attach it to a video with Combine › Add subtitle track or burn it in with Combine › Burn in subtitles. You can also translate it, shift timing, or convert it to WebVTT. You can also translate it or shift the timing, or convert it to WebVTT if you want to use it on a web page.