Step 01 · Drop
Your track, in one motion.
Drag a song onto the canvas. Hookline strips peaks, transcribes the lyrics with Whisper-turbo, and indexes every word's start and end — usually inside a minute.
drop a song
mp3 · wav · m4a · 320kbps
An AI lyric-video studio that lives in your browser — no servers, no subscriptions, no surprises.
Your song,scored tomotion.
Drop a track. We transcribe every word, pick a vibe that fits, and animate the lyrics over a backdrop tuned to your sound — then export an MP4 without ever leaving the tab.
Transcribed · 142 words
Vibe · cosmic
Render · 0:54 / mp4
9:41
cosmic
we fall into stars tonight
hookline · auto
How it works
Three steps. No timeline until you want one.
Hookline runs a deterministic five-stage pipeline behind the scenes — you just see the result of each beat as it lands.
Step 02 · Direct
Gemini picks the vibe.
A closed-set planner reads the lyrics and the song, then chooses one of seven moods — and a per-line display directive for emphasis, karaoke, or full-line reveals.
cosmic
sunset
noir
vapor
minimal
dreamy
smoky
Step 03 · Render
MP4, on your machine.
Remotion composes the scene clip-by-clip in the page itself. No queues, no Vercel timeouts, no exports lost to a slow function. Save and ship.
Transcribe
done
Compose
done
Encode
87%
MP4
2.1m
The seven vibes
A mood for every three-minute moment.
v1 ships locked to seven backdrops, each with its own typography, motion preset, and color palette. v1.5 opens the door to image and video backdrops.
01
expansive
we fall into stars
cosmic
02
warm
all the gold of july
sunset
03
cinematic
streets we knew by name
noir
04
nostalgic
a summer on tape
vaporwave
05
restrained
just a word for it
minimal
06
soft
in the blue hour
dreamy
07
melancholy
smoke and the silence
smoky sad
Why Hookline
Built on the quiet bits other tools skip.
Local-first, validated end-to-end, and free to use with your own keys. Three opinions we won't trade away.
— Local-first
Renders inside the tab.
Remotion composes every frame in the browser. Your audio never leaves your machine for encoding — Vercel's free tier doesn't get a chance to time out on you.
runtimebrowser / wasm
limityour hardware
cost$0 per minute
— BYOK
Your keys, your budget.
Bring a Gemini key and a Groq key. They live in browser sessionStorage, never on a server, never in git. You pay Google and Groq directly — we never mark anything up.
geminiflash-lite → flash
groqwhisper-large-v3-turbo
storagesessionStorage
— Validated
Every AI call is typed.
One Zod schema runs from planner output to editor state to the Remotion render. If a model hallucinates an emoji glyph or a motion preset, the schema rejects it before it can break your video.
schemav6 · single contract
retryflash-lite → flash
fallbackdeterministic
FAQ
A few quick ones.
Pennies. Whisper-turbo runs around a cent per song; Gemini Flash-Lite less. The render itself is free — it happens in your browser. We charge nothing on top.
No. Audio peaks, transcription, composition, and encode all happen in your tab. Transcript bytes pass through Groq for one call; nothing is logged on our side.
Yes — paste a clean lyric sheet and Hookline aligns it to the audio instead of re-transcribing. Useful when the singer mumbles or the mix is dense.
v1 is tuned for tracks up to about six minutes. Longer pieces work, but the browser encode gets heavy. v1.5 will chunk longer renders.
v1.5 unlocks image and video backdrops. v2 expands the format to reels, weddings, and podcast clips — same pipeline, new templates.
— now in open beta —
Drop a song. Watch it sing.
No signup wall. No download. Bring a Gemini key and a Groq key and you'll have an MP4 in about two minutes.