What are auto captions?
Auto captions are captions generated from a video's speech, with no one typing them. The audio is transcribed, every word gets its timing and the text is split into blocks that come in and out with the speech. Then you review names and terms and publish them as a file or burn them into the video.
Also called: automatic captions, auto-generated captions
From speech to captions
Word-level timing is what lets the sentence break into short blocks that appear the moment they are spoken.
| Step | What happens |
|---|---|
| Transcription | Speech becomes text, word by word |
| Timing | Every word gets a start and an end |
| Blocks | The text is split to the rhythm of the speech |
| Review | You fix names, terms and acronyms |
| Delivery | An SRT or captions burned into the video |
Why the review still matters
Captioning an hour of video by hand takes hours. Auto captions do the heavy lifting in minutes and leave you the part only someone who knows the content can do: the guest's name, the company acronym, the Bible book.
What separates good auto captions from bad ones is timing. If timing comes in blocks of several seconds, the captions lag. With the timing of every word, they follow the speech.
Auto captions in FilmScribe
We transcribe in 90+ languages with the timing of every word. You fix anything right in the transcript, and the fix carries over to your clips' captions and the SRT. Download the SRT on every plan, or burn the captions into your clip with one of 235 animated styles.
Auto and animated captions Captions that follow the speech, burned into your clip or as SRT, with 235 styles and 54 fonts. See the captions →Reviewing auto captions fast
- 01
Search for the hard words
Before you read, list the names, places, brands and jargon in the video. Search the transcript for each one and fix every spelling at once, which is faster than reading line by line.
- 02
Listen where the audio is weak
Errors cluster where people talk over each other, mumble, laugh, step away from the mic or speak under loud music. Jump to those moments and check them by ear.
- 03
Check numbers and sound-alikes
Speech recognition mixes up words that sound alike, like “their” and “there” or “fifteen” and “fifty”. Check every number, date and price against the audio, since a wrong figure misleads more than a typo.
- 04
Scrub the silent stretches
Long silences, music and noise sometimes produce captions with words nobody said. Skim the quiet parts of the video and delete any caption that appears while no one is speaking.
Related terms
See all terms →Frequently asked questions
Are YouTube's auto captions good enough?
They work for people watching on YouTube, but they stay there, with no styling and with whatever errors nobody reviewed. To take captions to other platforms or to your editor, you need a reviewed file or burned-in captions.
Do auto captions work in other languages?
Yes. In FilmScribe, the transcription detects the language on its own among 90+ languages, with English, Spanish and Portuguese the most used.
Do I need to review auto captions?
A quick read is always worth it. Names, technical terms and acronyms are the most common errors, and one fix in the transcript solves it everywhere.
Captions, clips and paper edits. All in one place.
Upload a video or paste a YouTube link. We transcribe every word with its timing, and you take the result to social media or to your editor.
See the result on your own recording before adding a card.