Automatic captions for Reels: how to make them and which ones people read
Most videos in the feed are watched without sound, so captions in a talking Reel aren’t decoration — they’re the main way to get your point across. Here’s what they should look like so people actually read them, and how to make them automatically without placing every word by hand.
- Captions appear word by word, in sync with your speech
- 7 styles: from a single line to “one word at a time”
- 8 colors + your own color or a color picked from a photo
- The line avoids your face, the title and graphics
- Text can be fixed before building and in the editor
Why Reels need captions
People scroll the feed on public transport, in a queue, next to a sleeping baby — far from everyone watches with sound. If a talking video has no text, the viewer only sees moving lips and leaves. Captions solve this and also set the rhythm: the eye catches on words as they appear.
Captions people actually read
- Short. 2–4 words on screen at once. Nobody reads a paragraph at the bottom of the frame.
- In time with speech. A word appears when it’s spoken. Running ahead or lagging behind is more annoying than no text at all.
- Large and high-contrast. White or light text, with a soft shade under it instead of a heavy outline.
- With an accent. One key word in color or larger — so the idea lands at a glance.
- Not on the face and not under the buttons. In Reels, the bottom 20% of the frame is covered by the description and buttons — text should sit higher, but not on the face.
How VARIA makes captions
Speech is recognized with the exact timing of each word, and sentences are split by meaning: a preposition stays with its word, “not” stays with the verb, a number with what it counts. Then the chosen style lays out the line:
Color — from a ready-made palette (olive, blue, light blue, powder, sand, burgundy, orange, purple, white), your own, or picked with an eyedropper from a photo of your brand. If the background behind the text is too light, the highlighted word’s color automatically gets lighter so it stays readable. Size, height and alignment (left, center, right) are set with buttons, and the preview instantly shows how the captions move.
Placement that avoids your face
With “Auto”, the edit finds the face in the frame and places the line so it doesn’t cover it: usually on the chest or above the head. When graphics or a title appear in the frame, captions move out of their way and return to their place right after. If the source already has burned-in text — a lower third, a label — our captions will avoid it too.
Mistakes that make people skip captions
- Text at the bottom of the frame. It’s covered by the Reels description and buttons — the viewer sees half the line.
- Five colors and three fonts. An accent works when there’s only one. One colored word per phrase is enough.
- A thick black outline. Looks cheap and eats the letters on a small screen. A soft shade under the text reads better.
- Text over the face. The viewer’s eye goes to the face — if there are letters there, they see neither the face nor the text.
- Filler words in the text. “Well, like, um” looks worse on screen than in speech. If a word is cut from the audio, it shouldn’t be in the captions either.
Captions help more than just people watching without sound: they’re read by people with hearing loss and by those for whom the video’s language isn’t their native one. That widens your audience with no extra effort.
Recognition errors and how to fix them
Speech recognition is very accurate today, but not perfect: surnames, brands, rare terms and fast speech sometimes come out with mistakes. That’s why you can open the text as a list and fix it before building, and afterwards — open the video in the editor. Filler words cut from the video disappear from the captions too — the text matches what you hear.
Captions only or a full edit
If you only need captions, choose the “Captions only” preset: the speech is cleaned of pauses, text appears, and nothing else. If you want the whole video designed — “Captions & title” or “Full edit”.
Pricing
Auto-edit is paid in packs: 5 edits — $7.05; 10 edits — $13.03; 25 edits — $29.77; 50 edits — $53.69. One edit is one finished video from a clip up to 3 minutes. Your first video can be a trial — with a watermark. Reposts are separate, from $5.98 for 30 reposts; 2 reposts free to start. Packs never expire.
FAQ
Can I make captions without the rest of the editing?
Yes, the “Captions only” style: pauses are cut, captions appear, no title or graphics. You can turn off “Smart cut” — then the video length won’t change.
Can I use my own font?
Anyone can change color, size, position and alignment. Each style already comes with its own font that supports Cyrillic; uploading your own fonts isn’t available to everyone yet.
Are captions burned into the video?
Yes, they become part of the picture — the video is posted as is, without a separate caption file.
What if a word is recognized wrong?
You can fix the text as a list before building, or afterwards in the editor. Rebuilding a paid video is free.
Does it work for English speech?
Yes, English speech is recognized too. Auto-editing works best with Russian.