How do you transcribe a video to text for free?

Start with the free route that needs no transcription at all. If the video already has captions, copy the transcript. If it does not, you can transcribe the audio yourself with open-source software on your own computer, or with a cloud service's free allowance. Each route has a cost in time, computer power or usage limits, and each needs a check against the video at the end.

Copy the transcript the video already shows

When a YouTube video has a transcript on its page, reading and copying it is the cheapest route. YouTube's terms allow you to view or listen to its content for your personal, non-commercial use ¹, and in our reading of the terms, copying text you can already read on the page stays within that permission. The text came from captions, so it has the errors that captions have. Read it against the audio before you rely on it. Our question on how to get a YouTube transcript explains where the transcript sits, and why the YouTube transcript is not showing covers the cases where none exists.

Whisper is free to use on your own computer

Whisper is the open speech recognition model from OpenAI, and it is free. The OpenAI repository states that its code and model weights use the MIT License ². A permissive license means you can use the software at no charge, and the license file sets the terms that apply to your copy. The cost is your time and your computer. Whisper runs locally, so you must install it and run it yourself, and you need enough memory for the model you choose. Our guide to transcribing with Whisper explains the setup.

Whisper can also invent text, including whole sentences where the audio has none ³. Check the draft for text that appears where the video has no speech, and for sentences that repeat.

Buzz gives you the same engine in a desktop app

Buzz is a free desktop app that runs Whisper on a personal computer, and its license is the MIT License ⁴. Its own documentation describes it as transcribing and translating audio offline on a personal computer ⁵. Its README lists live transcription from a microphone, and transcription of audio and video files and YouTube links ⁶. The YouTube link option is the step to check. Using it means getting the video's audio, and YouTube's terms bar downloading any part of its content unless the service permits it ⁷. Use a YouTube link in Buzz only for a video you have permission to download, or instead import an audio file you already have.

A cloud allowance is free only for a small amount

Cloud services often give you a free allowance. Google Cloud's V1 speech price table lists 0 to 60 minutes of standard recognition each month as free, per account ⁸. That is enough for a few short videos and not for a steady flow. Allowances and prices change, so check the current terms before you plan a project around one. Cloud services process your audio on their servers, which matters for private recordings.

Live captions give a free read, but the text may not be saved

If you want to follow a video without saving a transcript, live captions are free on the devices that have them. Chrome's Live Caption, Google says, works offline, including for audio and video files saved on the computer ⁹. The text is for reading, so check whether the app lets you copy or save it before you rely on the output. The guide to transcribing video lists the live caption options for Mac, Windows and Chrome.

Which free route fits your video

A free route can give you a usable draft for a short video, but you must check it against the audio. If the transcript is for a published or legal use, a human-checked transcript may be the better choice. Our question on how to transcribe audio for free covers the same options for recordings you already have.

What to do next

1. Look for a transcript on the video page. If it exists, copy it and read it against the audio.
2. If there is none, and you have the audio, install Whisper or Buzz and transcribe a one-minute sample first.
3. Check names, numbers and any silent or musical stretches, because those are where errors cluster.
4. Save the result as a text file, and add timestamps or captions only if you need them.

For what to add to a draft to make captions, see captions and subtitles. The overview of the free options for every kind of audio is in our overview of transcription.