What is forced alignment?
Checked against the product on October 22, 2026
Forced alignment is a technique that takes a known text and an audio recording and finds when each word is spoken. It is used to time captions precisely, for example after a transcript has been corrected by hand. It works best with clear audio and text that really matches what was said, and it struggles when words differ from the speech.
Definition
Ordinary transcription listens to audio and guesses the words. Alignment starts from the words and finds where they occur in the audio. This is useful when you have a trusted text, such as a script or a corrected transcript, and want captions to appear exactly in step with it.
Transcription versus alignment
| Aspect | Transcription | Forced alignment |
|---|---|---|
| Input | Audio | Audio and text |
| Output | Text with timings | Timings for known text |
| Risk | Misheard words | Text that differs from speech |
| Best when | No text exists | A reliable text exists |
Using it well
- Prepare a faithful textMatches what is said, including repeats.
- Use clear audioReduce noise and echo.
- Run alignmentMatch words to time.
- Check a few placesStart, middle and end.
- Fix driftIf the text differs, correct it.
- Export captionsReview them on a phone.
Common problems
- A script that the speaker did not follow.
- Missing or added words.
- Background noise and music.
- Different spellings of numbers and names.
- Several speakers talking at once.
Why it matters for translations
A translated caption cannot be timed from the audio directly, because the words differ. Timings are usually carried over from the original segments and then adjusted for length, which is why translated lines should be checked on screen.
How Averoai.pro relates
Averoai.pro builds captions from the speech and lets you correct the text in the caption editor, with captions timed to the audio. If you translate subtitles into around 70 languages, review the result on screen, because line lengths and timing can differ from the original.
A worked example
An editor corrected a transcript to fix names and jargon. The captions kept their timing, and a quick scan showed one segment drifting by a second where a sentence had been reworded. Matching the text to what was spoken fixed it.
Worth remembering
Alignment is about matching text to sound. A faithful text and clean audio make it accurate, and a text that departs from speech makes it drift. Always scan the captions at a few points before publishing. If you reuse the same corrected transcript for several clips, the timing work is done once, so it pays to fix the text carefully at the start rather than patching each clip separately later, when a small mismatch is easy to overlook.
Frequently asked questions
What is forced alignment?
Finding when known words are spoken in audio.
When is it useful?
When you have a reliable text and want exact timings.
Why does it drift?
The text differs from what was said.
Do translated captions need checking?
Yes, on screen.
Features that help with this
Related reading
- What are headroom and lead room?
Headroom is the space above a head, lead room the space in front of where someone looks. Learn why both matter, especially when you crop to vertical.
- What is eye line?
Eye line is where a speaker looks. Learn when to look at the lens, when to look away, how it affects trust and how to avoid reading from a screen.
- Dubbing vs subtitling: which suits your clips?
Dubbing replaces the voice; subtitling adds translated text. Compare cost, speed, feel and risks, and see which fits short clips for a new audience.
- What is text-to-speech?
Text-to-speech turns written words into synthetic voice; captions turn speech into written words. Learn the difference, uses and cautions, including disclosure.
- What is a video transcript?
A transcript is the written text of what is said in a video. See how it differs from captions and how it powers clip finding, captions and translation.
- Captions vs subtitles: what is the difference?
Captions and subtitles are often confused. Learn what each is for, how open and closed captions differ and which one short-form video usually needs.
Also available in Français