2 word-timed caption families every karaoke hook should use in 2026

Of nine caption families (45 presets), only Snap (Anton) and Reveal (TikTok Sans) are word-timed; the other seven land the whole line as a subtitle block — bookmark /caption-styles/group/snap and /caption-styles/group/reveal as one karaoke index, not nine twin posts.

Karaoke is a clock. Each spoken word has to hit. The catalog is nine groups of five, typed in src/lib/captionStyles.ts off src/data/caption-presets.json. kind is either timed or subtitle. Timed is Snap and Reveal. Everything else is a block. Do not restyle Classic until it “feels like Snap.” It will not.

Nine families, two clocks

Caption styles is the index of all 45. Group order in the loader is Classic, Paper, Ink, Halo, Hype, Signature, Headline, Snap, Reveal. Five variants each. getCaptionGroupSlugs() is those nine tokens.

Seven groups set kind=subtitle. The whole line arrives together. There is no effect field, no per-word scale, no typewriter. That is a phrase. It is not karaoke.

• Classic — TikTok Sans, white fill, black stroke, bottom. The everyday block. Hub: /caption-styles/group/classic.
• Paper — Montserrat, black on a white card.
• Ink — Inter, white on a black card.
• Halo — Poppins Bold, heavy ring.
• Hype — Oswald, condensed caps, coloured stroke.
• Signature — Pacifico, cursive, middle.
• Headline — Bebas Neue, tall caps, top. A title row, not a transcript.

If nobody spoke, neither timed group has a word to time. Overlay a CTA. If you wrote a line nobody said, timed overlays are authored beats, not karaoke. Add captions to a video is the transcribe-and-burn path: speech in, styled type out. Apply Snap or Reveal on that pass.

1. Snap — Anton, pop, one word punches

Snap is kind=timed, effect=pop, Anton. Each spoken word scales up and jumps to an accent. The punch is the current token. The rest of the line is not a construction.

Five presets, all bottom, all Anton, all pop. The base is word_pop: white fill, black outline, accent #FF4785. Variants change only the accent: Snap Gold #FFD700, Snap Cyan #22D3EE, Snap Lime #A3E635, Snap Mono #FFFFFF (scale only). There is no top variant. If the mouth sits in the lower third, the scale will hit the mouth.

Use Snap on a fast-talking hook, an ad read, comedic timing. Against a slow voiceover it looks frantic. Mono is the quiet Snap: scale without a colour jump, when the plate is already loud. Anton is a display face in the overlay registry, not a generate. Do not run Snap on a silent product plate. Do not ask Veo to “caption in Snap.” Veo will invent glyphs. Snap is a preset on speech.

2. Reveal — TikTok Sans, typewriter, the line builds

Reveal is kind=timed, effect=typewriter, TikTok Sans. Words appear one at a time and stay. The viewer watches a sentence get built. That is the other karaoke family. It is not Snap with a different font.

Five presets, all bottom, all TikTok Sans, all typewriter. The base is typewriter: white fill, black outline, live word accent #FFD700. Variants: Reveal Cyan #22D3EE, Reveal Pink #FF4785, Reveal Lime #A3E635, Reveal Mono #FFFFFF. Lime is not a new family.

Use Reveal when the voiceover is walking toward a punchline, a list that should be read as it arrives, a “wait for it” structure. Against a machine-gun hook it lags. Alignment has to be clean. A typewriter caption that lands a beat after the mouth looks broken in a way a static subtitle does not.

Do not treat Reveal as “Classic with a typewriter.” Classic is kind=subtitle on the same TikTok Sans. Subtitle families forgive a small drift because the eye reads a phrase. Reveal puts a single arriving word in charge. If that word is late, the build is the bug. Mono is the quiet Reveal: same typewriter, no gold flash.

Bookmark the karaoke index, not nine twins

Snap and Reveal are one job with two effects: pop versus typewriter. Bookmark the two group hubs. That is the karaoke index. The other seven groups are subtitle blocks. They already have family pages. Do not write nine “karaoke” twins as if kind were a theme.

House style is one timed family plus one accent, written down. Snap Gold, bottom. Reveal Mono, bottom. Switching to Hype on Thursday because the last Reel felt loud throws away the recognition the word clock already earned. Hype is a subtitle block. It will never pop a token.

Timed overlays are a different clock. You write every string and set start_sec / end_sec. Nothing is transcribed. Use overlays when nobody said the line. Use Snap or Reveal when somebody said the line and you want each word to hit.

Burn after crop. A bottom karaoke that sat clean on 16:9 sits in platform chrome after a 9:16 reframe. Picture lock, ratio lock, then Snap or Reveal. If someone spoke and each word needs its own beat, this index. If the whole line can land as a phrase, Classic. If you wrote a hook nobody said, overlay. Karaoke is two families. The other seven are a different object.

FAQ

Can Classic karaoke if I pick a punchy font?

No. Classic, Paper, Ink, Halo, Hype, Signature, and Headline are kind=subtitle. They do not run effect=pop or effect=typewriter. Snap and Reveal are the word-timed groups. Font choice on a block does not become karaoke.

Is a timed text overlay the same as Snap or Reveal?

No. Timed overlays are authored lines with timestamps you typed. They do not karaoke. Snap and Reveal are timed caption styles on transcribed speech. Overlays do not scale or typewriter each spoken word. Snap and Reveal do not invent a line nobody said.

What if the take has no speech?

Do not use Snap or Reveal. Overlay one authored line, or use timed overlays for beats you wrote. A timed family with nothing spoken is a style waiting on a transcript that does not exist.

Snap or Reveal for a fast hook?

Snap. Pop matches a fast-talking hook; Reveal will look late even when the transcript is perfect. Reveal is the build: punchline, list, “wait for it.” Fast talker, Snap. Building a line, Reveal. Bookmark both hubs. Do not maintain nine karaoke posts.