How to Make Hormozi-Style Captions Without a Subscription

Hormozi-style captions are large, all-caps words in a heavy sans-serif, white with a thick black outline, shown one to three at a time and synced word by word, with a single highlight colour — usually yellow — on the word being spoken. You can make them with an online caption tool, a free editor such as CapCut, a desktop app with a ready-made preset, or the Whisper command line plus FFmpeg.

The part people underestimate is the timing. The look is easy to copy in a style panel; what makes it work is that every word appears exactly when it is said. That requires word-level timestamps, and it is the reason most of this article is about where those come from.

What makes a caption "Hormozi-style"?

Alex Hormozi's short-form clips made one caption treatment so recognisable that people now search for it by name. Stripped down, it is six decisions:

1. A very heavy sans-serif. Montserrat Black or ExtraBold is the usual match; Anton and Archivo Black are close. Weight matters more than the exact face.
2. All caps. Capitals give every word the same height, which makes a two-word caption read as one shape.
3. White text, thick black outline. The outline is what makes it readable over any footage, including a white wall or a bright shirt.
4. One highlight colour. Yellow around #FFD93D is the default. It marks either the word being spoken or the one word that carries the meaning.
5. One to three words on screen. Two is typical. The viewer catches words instead of reading lines.
6. Centre-low position. Around the lower third of a 9:16 frame, above the caption and button area that TikTok, Reels and Shorts draw over the video.

There is often a small pop-in animation as each group appears. It is the least important part of the style and the first thing to drop if your tool does not do it well.

Why this style works on muted video

Most short-form video is watched with the sound off, and captions are the only way a muted viewer follows speech. That part is not Hormozi-specific — any captions help.

What this treatment adds is pacing. A full subtitle line asks the viewer to read at their own speed, and they fall behind or look away. One or two words at a time, timed to the voice, forces the eye to follow the speaker's rhythm instead. It also sidesteps the reading-speed problem that long lines have, because no single caption holds enough characters to overload anyone.

The trade-off is that it is loud. On a calm interview, a documentary or a tutorial where the screen matters, it covers the picture and fights the content. It belongs on talking-head clips built to stop a scroll.

Four ways to make Hormozi captions

Online tools such as Submagic are the fastest route to the look: upload, pick a template, export. They are also subscriptions with length caps — the Submagic comparison has the current numbers — and the video has to leave your machine.

CapCut is the default free answer for creators. Its auto-captions support the word-by-word style and it runs on a phone, which none of the desktop options do.

A desktop app with a preset keeps the file local and removes the length question. The rest of this article shows the workflow in Sablate, because that is the tool we build, followed by the fully manual route.

Whisper plus FFmpeg costs nothing and gives you complete control, at the price of writing commands and, for the highlight, editing a subtitle file by hand.

Making them in Sablate

Sablate is a desktop app for Windows and macOS that transcribes locally with Whisper, which returns a timestamp for every word. That is the input the highlight needs, and nothing has to be uploaded to get it.

1. Open the video and transcribe. The Balanced model is enough for clear speech in a quiet room; use Accurate if there is music under the voice.
2. Fix the text. Add recurring names and brand terms to the custom vocabulary once, so they come out right on the next video too.
3. Rewrap to short captions. The editor's line-splitting tool rewraps the whole track at one, two, three or four words per caption.
4. Pick the Hormozi preset. It is one of thirteen built-in presets, in the Viral group alongside Beast, Reels, One Word, Karaoke and Pop. The word highlight can be a colour change, a box behind the word, a left-to-right fill, or one word at a time.
5. Set the frame to 9:16. A horizontal source can be cropped or filled with a blur of the picture, and the captions are laid out for the vertical frame rather than cropped with it.
6. Render a 15-second sample first, then the full video.

The render uses libass, the same subtitle engine as the preview, so the font, the outline and the highlight in the exported .mp4 match what you saw in the editor. The fonts that suit this style — Montserrat ExtraBold, Anton, Archivo Black, Bebas Neue and Bangers — ship with the app, so the render never quietly falls back to a different face.

On the free plan you can preview every preset and burn in videos up to ten minutes, with a small corner mark. The word-by-word karaoke highlight, custom presets and an unmarked render are part of Pro, which is $29 once.

Making them with Whisper and FFmpeg

The manual route has two steps: get word-timed captions out of Whisper, then burn them in with FFmpeg.

OpenAI's Whisper command line can cap every caption at a fixed number of words:

whisper clip.mp4 --model small --word_timestamps True --max_words_per_line 2 --output_format srt

That gives you an .srt with two words per caption. FFmpeg can then burn it in with the Hormozi look forced onto it:

ffmpeg -i clip.mp4 -vf "subtitles=clip.srt:force_style='FontName=Montserrat Black,FontSize=22,PrimaryColour=&H00FFFFFF,OutlineColour=&H00000000,BorderStyle=1,Outline=3,Alignment=2,MarginV=60'" -c:a copy out.mp4

Three things trip people up here:

• The font must be installed, or libass silently substitutes another one and the whole look changes.
• Colours are written backwards. Subtitle colours in this syntax are &HAABBGGRR — alpha, blue, green, red — so yellow #FFD93D becomes &H003DD9FF.
• A plain .srt cannot highlight one word. The yellow word needs an .ass file with an inline colour override on that word, like this:

Dialogue: 0,0:00:01.20,0:00:01.85,Default,,0,0,0,,{\c&H3DD9FF&}NOBODY{\c} CARES

For a one-minute clip that is manageable by hand. For a twenty-minute video it is an afternoon, which is the point at which a tool with a preset starts paying for itself. If you are new to the format, the SRT vs VTT vs ASS guide explains why only .ass can carry this styling.

Getting the details right

Keep the captions out of the interface. TikTok, Reels and Shorts all draw buttons down the right-hand side and a description over the bottom of the frame. Captions that sit lower than roughly the bottom fifth get covered.

Highlight meaning, not every word. A sweep that lights up each word in turn is the karaoke variant. Hormozi's own clips often highlight only the word that carries the point. Both work; mixing them in one video does not.

Break captions at natural phrases. "NOBODY CARES / ABOUT YOUR / PRODUCT" reads better than "NOBODY / CARES ABOUT / YOUR PRODUCT", even though both are two words per caption. Fix the worst breaks by hand.

Check on a phone before publishing. A caption size that looks right on a monitor is often too large on a phone held at arm's length, or too close to the edge on a narrow screen.

Make it once in each language you need. If the same clip goes out in Spanish and Portuguese, transcribe once and translate the text onto the same timing — the multilingual workflow covers why that beats re-transcribing per language.

The short version

Hormozi-style captions are heavy all-caps words, white with a black outline, one to three at a time, with one highlight colour, timed word by word. The style is easy to copy; the word timing is the part that takes a tool.

If your clips are short and you work on your phone, CapCut is the simplest free route. If you want the file to stay on your machine, need long videos, or publish the same clip in several languages, download Sablate for Windows or Mac and try the Hormozi preset on a real clip. Whichever tool you use, burn in at the final aspect ratio rather than cropping afterwards.