How to Transcribe a Video
Four routes, what each one actually costs you, and when each is the right answer. Written from what this tool can and cannot do, not from a feature list.
Route 1: use the captions the platform already made
YouTube, Zoom and Teams all generate captions. If the recording is on one of those platforms and you only need the words, open the transcript panel and export it. It costs nothing, takes seconds, and the timing data is already there.
Limits: punctuation is poor, names are wrong, and there is no transcript at all if the uploader turned captions off.
Route 2: transcribe the file in a browser tool
Drop the audio or video file into a page like this one. Your browser decodes it, uploads short slices, and returns text. Nothing to install, works on a work laptop.
Limits: browser support decides which containers work, and long files are processed slice by slice. Start with the free 2-minute preview before paying anything anywhere.
Route 3: run a speech model locally
Whisper and whisper.cpp run on your own machine. Nothing leaves the computer, there is no per-minute cost, and you can batch a folder of files overnight.
Limits: a real install (Python or a compiled binary), a decent CPU or GPU, and you own the troubleshooting. This is the right route if you transcribe a lot and care about data never leaving your machine.
Route 4: a monthly transcription service
Upload and get transcripts with speaker labels, timestamps and an editor. The most polished output, and the most expensive way to transcribe occasionally.
Limits: subscriptions renew whether you use them or not, and per-minute allowances are easy to overrun.
Side by side
| Route | Cost | Privacy | Best for |
|---|---|---|---|
| Platform captions | Free | Already on the platform | Text you only need to read |
| Browser tool (this site) | $4.99 one-time per file | Local decode, slices uploaded | Occasional files, no subscription |
| Local model | Free, your hardware | Nothing leaves the machine | Volume work, sensitive audio |
| Monthly service | Subscription | Uploaded to a vendor | Teams needing editor features |
Do this before you pay anyone
- Transcribe the first two minutes of your actual file. Every route looks good on clean studio audio; yours may have crosstalk, music or an accent.
- Read the output for names and numbers — that is where recognisers fail first.
- Only then decide between one-time and subscription pricing, based on how many files you really have.
FAQ
Which route is cheapest?
YouTube's own captions, when they exist and are decent — free and instant. Transcribing here costs $4.99 per file with no subscription, which is cheaper than a monthly plan if you transcribe occasionally.
Which route is most private?
A local model (Whisper or whisper.cpp on your own machine) keeps everything offline. This site is the middle option: local decoding, short slices uploaded.
Which route is fastest for one long file?
A paid cloud service usually finishes a one-hour file in a few minutes; this site depends on your upload speed for the slices. Speed is not why you would choose it.
Try it on your file
No account needed for the preview. Video to transcript · MP3 to text · Voice recording to transcript