To transcribe a video, you transcribe its sound track: the speech is turned into text and the picture is not read at all. So words that only appear on screen, such as slide titles, will not be in the text. There are three ways to do it: upload the whole video to a service, upload just its sound, or run the recognition on the computer where the file is. Dalm is a Mac app that does the third on the Free plan, for videos up to four hours: the speech is recognised on your Mac and no audio is sent anywhere, and the text is synced to our servers.
Only the sound track is read, and what that leaves out
A video is a picture and a sound track, and transcription works on the sound. Whatever is only shown — slides, captions burned into the picture, a name under a speaker’s face — is not in the text, and neither is anything that was shown but never said. A video with no speech in it, such as music over a title card, comes back with “We could not make out a single word.”; a file with no sound track at all is refused. What decides is what is inside the file, not the letters after the dot.
If the slides matter, keep them next to the transcript, or say the main points aloud while you record.
Upload the video, upload only its sound, or keep the file where it is
| Upload the whole video | Upload only its sound | Keep the file where it is | |
|---|---|---|---|
| What travels | The whole file, picture included, to the service | A sound file you make first, to the service | The video stays in its folder and no audio is sent anywhere; in Dalm the text is synced to our servers |
| Before the text starts | The upload has to finish | You take the sound out, then upload a much smaller file | The app takes the sound out of the video itself |
| What it asks of you | A browser, and a file within the service’s limits | A tool that takes the sound out of a video | A Mac with Apple Silicon · macOS 15.0 or later, an account, and a model download on first launch |
| What to check | How large and how long one file may be, and how long it is kept | The same questions, for the smaller file | Whether the text leaves too — in Dalm it does |
Taking the sound out and uploading only that is an honest way to send less to any service, and it does not need our app at all. If what matters more is that the recording does not leave your computer at all, that is a separate question — whether a recording should leave your computer at all.
Before you start: the kind of video and how long it runs
- A call saved as a video file. It goes in as a meeting with several voices.
- A screen recording of a call. Whether the other people’s voices are in it was decided when you recorded — see what the Mac’s own screen recording captures.
- A webinar or a lecture filmed on camera. Mostly one voice. Bring it in as a meeting if you want a summary — one speaker is enough; as a dictation it is kept as text only. What a student gets from a recorded lecture is on lecture transcription for students.
- A video you made yourself. Start from your original file rather than a copy you downloaded back.
One file can run from 0.3 seconds to four hours, and how large it is plays no part; a longer recording has to be cut into parts before it goes in. Whether you choose dictation or a meeting decides where the text is filed, and a meeting also gets a summary with quotes. The choice is made once: a dictation cannot be turned into a meeting afterwards, so the file would have to come in again.
Creating a transcript from a video in Dalm
You bring the video in, and the app opens it in the folder where it is, without making a copy. While the sound is taken out of the video, the window says “Preparing”. The speech is then recognised on your Mac and no audio is sent anywhere; the text is synced to our servers. The temporary copy of the sound is erased as soon as the work is done, and you can close the window — the transcript appears in your list when it is ready. The steps for any file are under transcribing a file on the Free plan, and whether it needs a network has a page of its own.
What comes back, and what the Free plan covers
The transcript lands with your meetings or your dictations, depending on what you chose. Bringing in a video is part of the Free plan, which has no end date. Dalm needs a Mac that meets the system requirements and an account; while the app is not public, the way in is the waitlist. What the paid plans add, and what they cost, is on the pricing page.
Questions
- Is video transcription different from audio transcription?
- Only in the file. The speech in a video is a sound track like any other recording, and that track is what gets transcribed. What comes with it is different: a larger file, and a picture whose text, from slides to names on screen, does not end up in the transcript.
- Do I need to extract the audio before transcribing a video?
- For a service you upload to, taking the sound out first means sending a much smaller file. For Dalm, no: the app takes the sound out of the video itself, keeps that copy only while it works and erases it afterwards, and the video stays where it is. The speech is recognised on your Mac and no audio is sent anywhere; the text is synced to our servers.
- Can I transcribe an MP4 to text?
- Yes, if your Mac opens it as a video and it has a sound track. Dalm decides by what is inside the file rather than by the letters after the dot, so the same holds for other video files macOS can play.
- Will the text on my slides be in the transcript?
- No. Only what is said is transcribed, and the picture is not read. If the slides matter, keep them next to the transcript, or say the main points aloud while you record.
- How long can the video be?
- Anything from 0.3 seconds up to four hours in one file, whatever its size. A recording that runs longer needs to be cut into parts before it goes in.
- Can I transcribe a video from a link?
- The app works from a file on your Mac. If the video is one you published, start from the file you uploaded, as set out for your own published video.
- Can I transcribe a video for free?
- Yes: bringing in a video and getting its text is part of the Free plan, which has no end date. You need a Mac with Apple Silicon · macOS 15.0 or later and an account; while the app is not public, the way in is the waitlist.