What to expect from this guide: the steps for getting one usable video transcript from a long recording, in one upload. On VideoText that means Long Video Transcription: Pro accepts speech files up to 4 hours (240 minutes) each and uploads up to 10 GB, within 30 hours of speech a month. Free accepts files up to 30 minutes.
Most transcription problems with long recordings are not about accuracy. They come from the file being cut up to fit a limit, and from the transcript that comes back being too long to navigate.
Why splitting a long recording causes problems
When a tool cannot take the whole file, the usual workaround is to cut it into parts. Every cut costs you something:
| What breaks |
Why |
| Speaker labels |
Each part is labelled on its own, so "Speaker 1" in part two may be a different person from "Speaker 1" in part one |
| Timestamps |
Each part starts again at 00:00:00, so every time after the first cut has to be offset by hand |
| Sentences |
A cut lands mid-sentence and the words either side of it are dropped or repeated |
| Your time |
You merge the parts, renumber speakers and fix the boundaries before the transcript is usable |
The same limit catches people building their own pipeline: OpenAI's speech-to-text API accepts files up to 25 MB per request, so a long recording has to be compressed or chunked before it can be sent at all.
Processing the recording as one file avoids all four. That is the main thing to look for.
Step 1: Check the per-file limit, not the monthly one
Tools usually state two limits: how many minutes you get a month, and how long a single file can be. For a long recording, the second one decides whether it works.
- Free plans commonly stop at 30 to 45 minutes per file.
- Paid plans vary from 90 minutes to several hours per file.
- On VideoText, the length of your file is read in your browser before anything uploads, so you see straight away whether it fits your plan.
On Free, a file over 30 minutes shows the Pro option. If you upgrade from there, checkout opens in a new tab and your file stays selected, so you do not have to find a multi-gigabyte file again.
Step 2: Make the file small enough to upload
A 4-hour video in 1080p is often several gigabytes. Two ways to make it smaller:
- Upload the audio instead of the video. Transcription only uses the speech. The audio track of a 4-hour recording is a small fraction of the video, and MP3, M4A and WAV all work.
- Compress the video. Lowering the video bitrate does not change the audio. Compress Video handles files up to 2 hours on Pro; for longer files, extract the audio instead.
Step 3: Upload the whole recording and let it run
Pick the full file. Do not trim it unless you really only need part of it.
- If your connection drops, the upload resumes rather than starting again.
- The job runs on the server, so you can close the tab and come back to the result later.
- A long recording takes longer to process than a short clip. Plan to come back to it rather than watch it.
Step 4: Fix speakers once, then use chapters to navigate
A transcript of several hours is only useful if you can find things in it.
- Rename speakers once. Changing "Speaker 2" to a real name applies across the whole transcript, because the whole recording was labelled in one pass.
- Use chapters to jump around. The chapter list breaks hours of text into sections, so you can go straight to the part you need.
- Check names and terms, not every line. Proper nouns, product names and jargon are where speech recognition most often needs correcting.
Step 5: Export what the next step needs
| You need |
Export |
| Notes, minutes or a document to edit |
TXT or DOCX |
| Captions for the video |
SRT or VTT (the same timing as the transcript) |
| Show notes or a course outline |
The summary and chapter list |
| The transcript in another language |
Export SRT, then translate it with the timing kept |
Common questions
Is accuracy worse on long recordings? Length on its own is not the problem; audio quality is. Background noise, people talking over each other and distant microphones hurt accuracy at any length. A long recording just has more chances for those to happen, so check the noisy sections first.
What about recordings longer than 4 hours? Split them into as few parts as possible, ideally at natural breaks such as a pause or a change of session, and process each part as its own file.
Can I get subtitles for the same long video? Yes. Long Video Subtitles produces SRT or VTT for the full running time. To burn subtitles into the video, note that Burn Subtitles handles videos up to 2 hours on Pro; for longer videos, upload the subtitle file to your player instead.