How to Add and Edit AI Captions in CapCut: A Complete Guide

The first time I used AI captions in CapCut, I trusted them completely, and that was my mistake. I recorded a short tip video, tapped Auto Captions, exported it, and posted it without reading a single line. Later, a viewer commented that one word was completely wrong, and it changed the meaning of my sentence. It was a small word, but it made my tip sound confusing. Since then, I never post AI captions without reading them first. AI captions save a huge amount of time, but they still need a human eye at the end.

If you make Reels, Shorts, or TikTok videos, captions are no longer optional. Many viewers watch with the sound off, in public places, or while scrolling quickly. Good captions help them follow your video even without audio. They also help viewers who are hard of hearing, and they make your key points stand out.

In this guide, I will show you how to add and edit AI captions in CapCut, step by step, and how to style them so they are easy to read on a phone. You will also learn the mistakes I made, so you do not have to repeat them.

What are CapCut AI captions

CapCut AI captions, often called Auto Captions, turn the speech in your video into on-screen text automatically. CapCut listens to the audio, detects the words, and places each line on the timeline at roughly the right time.

This is very different from typing captions by hand. Typing captions for a one minute video can take 15 minutes or more. Auto captions can do the first draft in less than a minute, depending on the length of your video and your internet connection.

But auto captions are a first draft, not a final version. They can mishear names, numbers, accents, and words that sound similar. They can also split a sentence in an awkward place, or put a line on screen for too short a time to read. That is why the editing step matters as much as the generation step.

Before you start: record clear audio

AI captions can only be as good as your audio. If your audio is muffled, noisy, or very quiet, CapCut will guess more words, and more guesses will be wrong.

Record in a quiet room if you can. Speak clearly, but do not shout. Keep the phone close enough that your voice is clear, and avoid background music that is louder than your voice. If you are recording outside, try to face away from wind and traffic.

I learned this the hard way. I once recorded a tip on a busy street, and CapCut turned a simple sentence into something completely different because a passing car covered half of my words. The video was fine, but the captions were not. Now I always do a quick audio check before I start editing.

Step by step: how to add AI captions in CapCut

Step 1: Start a new project and add your video

Open CapCut and start a new project with the video you want to caption. Make sure the video has the audio you want to caption. If you added music over your voice, the captions may try to transcribe the music instead, so keep your voice track clear.

Step 2: Open Text and choose Auto Captions

Tap Text at the bottom of the editor, then look for Auto Captions. In some versions, it may be labelled as Auto Captions or Captions. Tap it, choose the language you spoke in the video, and start the generation.

Choosing the right language is important. If you spoke English but selected another language, the captions will be wrong from the start. If your video mixes two languages, choose the main language first, and be ready to fix the mixed parts by hand later.

Step 3: Wait for generation, then read every line

CapCut will take a few seconds to generate the captions. When they appear, do not export yet. Play the video from the start and read each caption line as it appears. This is the most important step in the whole process.

Look for wrong words, missing words, and names or numbers that were misheard. Also look for lines that are too long. On a phone screen, a caption with more than two lines is hard to read quickly. If a line is too long, you will split it in the next step.

Step 4: Edit the text

Tap on a caption line to edit it. Fix spelling, fix wrong words, and remove filler words if they make the sentence confusing. You do not need to keep every um and uh. Captions should match what you meant, not every sound you made.

You can also split a long caption into two shorter captions, or merge two very short captions into one, if that makes reading easier. The goal is simple: a viewer should be able to read each line in the time it is on screen, without pausing the video.

Step 5: Style the captions for readability

Now tap on a caption to open its style options. Choose a font that is bold and easy to read. Avoid thin, fancy fonts that look pretty in the editor but disappear on a small phone screen.

Choose a size that is large enough to read without zooming, but not so large that it covers your face or the main action. Choose a colour with strong contrast, such as white text with a dark background, or yellow text on a dark video. Many creators use a highlight or background box behind the text, because it keeps the text readable even when the video behind it is bright.

Place the captions in a safe zone. For Reels, avoid the very top and very bottom of the screen, where Instagram buttons and captions can cover your text. The middle-lower area usually works best, but always preview on a phone-sized screen before you export.

You can also add a simple animation, such as a word-by-word highlight, if CapCut offers it in your version. Keep animations subtle. A flashy animation on every word can distract from what you are actually saying.

Step 6: Check timing and sync

Play the video again and watch the timing. Each caption should appear just as you say the words, not a second early or late. If a caption appears late, viewers will read it after they have already heard you, and it will feel out of sync.

If CapCut allows you to drag a caption block on the timeline, adjust any block that feels off. Pay special attention to the first caption. If the first caption is late, viewers may think the video has no captions at all and scroll away.

This timing check is similar to beat sync work. In our guide on how to make a beat sync video in CapCut, we talk about cutting on the beat. Captions work the same way. Timing is not a small detail. It is what makes the video feel professional.

Step 7: Export and do a final phone check

Export at 1080p if that option is available. After export, open the video on your phone, not just in the editor. Watch it once with sound off. Can you follow the whole video from captions alone? If not, go back and fix the lines that were unclear.

Also watch it once with sound on. The captions should support your voice, not compete with it. If a caption is too big, too bright, or in the wrong place, fix it before you post. It is much easier to fix now than after viewers have already seen it.

Common AI caption mistakes to avoid

Trusting auto captions blindly is the biggest mistake. AI captions are fast, but they are not perfect. Always read them.

Using one long line for a whole sentence is another mistake. Split long sentences into short, readable chunks.

Placing captions over your face is also common. If your face is the main subject, move the captions lower or to a side where they do not cover your expression.

Using low contrast colours is another problem. Light grey text on a bright video is almost invisible. Use strong contrast every time.

Forgetting to check names, numbers, and brand names is also easy to miss. AI often mishears these first. If your video mentions a product, a price, or a date, check those words twice.

How to style captions for Reels

For Reels, keep captions short, bold, and centred or lower-centred. Use one or two lines at a time. Highlight one key word per line if that helps viewers follow your point, but do not highlight every word. Highlighting everything is the same as highlighting nothing.

Avoid using more than one font style in the same video. A consistent style makes your videos recognisable. If you change fonts every video, your profile can start to look messy, even if each video is good on its own.

If you use CapCut templates for Reels, check the template’s default caption style before you post. Some templates use a style that looks good in the preview but is hard to read on Reels. You can usually change the font, size, and position to match your own style.

A simple caption workflow you can repeat

Record clear audio. Generate auto captions. Read every line. Fix wrong words. Split long lines. Style for contrast and size. Check timing. Preview with sound off. Then export.

This workflow takes a few extra minutes, but it saves you from comments about wrong captions, and it makes your videos accessible to more viewers. Over time, it becomes fast. I can now caption a one minute video in about five minutes, including the reading check, because I follow the same steps every time.

Key takeaways

• CapCut AI captions turn speech into text automatically, but they are a first draft, not a final version.
• Always read every caption line before you export, especially names and numbers.
• Split long captions into short, readable lines of one or two lines each.
• Use bold, high-contrast text in a safe zone that does not cover your face.
• Check timing so captions appear as you speak, not early or late.
• Preview with sound off to make sure viewers can follow the video from captions alone.

Final thoughts

AI captions changed how I edit, because they removed the most boring part of the job. I no longer type every word by hand. But they did not remove the need to think. I still need to check the words, choose a readable style, and make sure the timing feels right.

Think of AI captions as a helpful assistant, not a replacement for you. Let CapCut do the first draft, then use your own judgment to make it clear, accurate, and easy to read. When captions are done well, viewers do not notice them as a separate feature. They just understand your video better, with or without sound. And that is exactly what good captions should do.