A tutorial is ready to subtitle when viewers can follow the instructions without guessing at product names, commands, or timing. A transcript is the starting point: review it before publishing the SRT file.
This guide describes the transcribevideototext editing workflow. The short subtitle below is an illustrative example, not a transcription benchmark or a claim about processing speed.
1. Start with the recording you will publish
Use your final edit where possible. Cutting an introduction after generating subtitles shifts everything that follows. Upload a supported video, such as MP4, MOV, or WebM, or choose Audio for an audio recording. Sign in before uploading so that completed results can be associated with your account.
The uploader accepts files up to 5 GB and holds up to 20 tasks in its queue. These limits do not mean that 20 recordings are transcribed simultaneously. Upload, queue, and processing time vary. For your first attempt, choose a short recording with clear speech.
Public-link imports depend on whether the source can actually be downloaded. If an import fails, use a recording you have permission to download and upload it directly. Whole-playlist transcription and automatic speaker separation are not currently available.
2. Review the text against your recording
Open a completed task from Saved transcripts. Uploaded recordings can use temporary private playback links. If playback is unavailable, select the original file in the editor for local playback; this selection does not upload the file again. Browsers do not play every media codec, even when a file can be transcribed.
Click a segment's timestamp to seek to that part of the recording. Check names, numbers, commands, and negations carefully. For a software tutorial, “enable” and “disable” can both be correctly spelled while only one matches the instructions.
Automatic edits save to your account after a short pause. Wait for Saved to your account before leaving. A connection error leaves changes in the current tab; retry saving or download a copy. If another tab saved a conflicting version, export your local changes before reloading. Demonstration transcripts are only retained during the current visit.
3. Correct repeated terms
Use Find exact text, enter the replacement, then choose Replace all. Replacement is case-sensitive and applies to the whole transcript, not just visible search matches.
For example, replace a consistently misspelled product name only after confirming that every occurrence has the same meaning. A global replacement is not a model glossary: it fixes the transcript after recognition and does not retrain the model or change future recordings.
4. Fix a constant timing offset
If every subtitle appears one second late, enter -1 in Shift all subtitles (seconds). If they appear one second early, enter 1. The adjustment applies to every segment and is saved with the transcript.
An offset that would move a segment before zero is rejected without changing the timestamps. Progressive drift needs a different repair: inspect the beginning, middle, and end of the recording and check that you are using the same edit. See our subtitle timing drift guide.
5. Export SRT or WebVTT
Choose an export line width, then download:
- TXT: plain text for notes and reuse.
- SRT: timestamped subtitles for many video editors and publishing platforms.
- VTT: WebVTT captions for compatible web players.
Line wrapping preserves cue timing. It does not automatically split a long segment into several shorter timed cues. Check long captions in your target editor and split them where needed.
Here is a manually written SRT example:
1
00:00:01,000 --> 00:00:04,000
Open the project settings.
2
00:00:04,500 --> 00:00:08,000
Choose the language spoken
in your recording.We have not used this sample to measure recognition accuracy or claim compatibility with a particular video editor version. Test the exported file in your own publishing workflow. Compare format details in TXT, SRT, and WebVTT explained.
A short pre-publication check
Play the beginning, middle, and end of the final video with its subtitles. Confirm the first caption does not arrive late, instructions use the correct names and numbers, and the final caption still matches the recording. Review any cue that is too long to read comfortably.
The current workflow exports transcripts and same-language subtitles. It does not translate subtitles, burn captions into a video, or generate a dubbed audio track.
Open the video transcription workspace or review the current plan limitations.
