Use the Lyric Video wizard and on-device Whisper transcription to get lyrics on screen at exactly the right moment, styled to stay readable, then export.
Before you start
A lyric video needs three things: a visual that moves with the music, lyrics that appear at the right moment, and text that stays readable on top of everything. Novus does all three in the browser — no video editor, no account, and your audio never leaves your device.
Open the Lyric Video Creator. Its three-step wizard — Pick Engine, Enter Lyrics, Style & Export — collects your choices and drops you into the editor with the Captions controls already open. For the backdrop, four modes are tuned to leave room for text: Waveform Line, Waveform Ribbon, Filled Wave and Line Spectrum. Upload your audio, and paste your lyrics one line per row (each row becomes one on-screen caption).
One line per row
A pasted paragraph becomes one giant caption. Break your lyrics where you want them to change on screen. No lyrics typed out? Click Transcribe Audio in Editor to auto-generate them.
With audio loaded, AI Generate transcribes your song using Whisper, running 100% in your browser — no API key, no upload. An AI model dropdown offers Fast (~75 MB), Balanced (~145 MB) and Accurate (~250 MB); pick Accurate for the best word timing on your final pass. Songs are harder than speech, so tick Isolate vocals first (needs WebGPU) to separate the voice from the mix before transcribing.
Treat it as a timed draft
Whisper returns word-level timing, but names, slang and ad-libs often need a quick text fix. The timing skeleton is usually solid — you clean up the words, not the sync.
Each caption is a row with Start and End fields (adjustable in 0.1s steps). If AI or imported lyrics run consistently early or late, do not retime every row — drag the Timing offset slider (−3 to +3s) to shift them all at once. Turn on Karaoke timing to highlight each word as it is sung (this is what makes it feel synced rather than subtitled). Do a full playthrough to catch lines that linger or overlap.
The Style selector offers ten presets (Clean White, Subtitle Bar, Neon Glow, Bold Impact, Cinematic, Karaoke, Pop Pill, Dark Minimal, Fire Text, Glass Card). Readability beats decoration: over a busy engine, pick Subtitle Bar or Dark Minimal so text sits on its own contrast layer, and keep one or two lines on screen at once. Use per-line overrides sparingly — a bigger chorus line reads as intentional; ten different styles read as noise.
Open Export Settings, pick your Shape (16:9 for YouTube, 9:16 for Shorts), quality and format, and export — your caption timing, karaoke highlighting and timing offset all render exactly as they preview. One extra button worth knowing: Export SRT downloads your lyrics as a standard subtitle file you can upload to YouTube as closed captions or keep as a portable backup of your timing work.
Test before the full render
Export a 15–20 second test of the first chorus before committing to a full 4K render — timing mistakes are far cheaper to catch early.
Ready to try it?
Open the editor and follow along — free, no account, watermark-free exports.
Open the Editor