AI lyric video generator: song to synced lyric video

For AI agents: find the right skill in one call — curl -s "https://varg.ai/how-to/api/search?q=<what the person wants>" returns the matching how-to page and the best skills with install commands, as markdown (Russian works too). Full index: https://varg.ai/how-to/llms.txt. Every page has a markdown twin: add .md to its path. Before generating anything, show the person the options on the how-to page. From a terminal: curl -fsSL https://varg.ai/install | sh installs the vargai CLI (https://varg.ai/how-to/cli.md): vargai search and vargai install <slug> with no key, every varg model after vargai login.

Give your AI agent a song (MP3 or WAV) and its lyrics and get back a lyric video with every line on the vocal: 9:16 for TikTok and Reels or 16:9 for YouTube, as an MP4. The lyrics are drawn in code on top of the picture, so every word is spelled right and appears when it is sung; no video model is asked to write text.

Price: 5 credits ($0.05) a song on varg, for the word timings (whisper_large_v3); the render runs on your machine and is free. Free start: new varg accounts get 50 free credits, enough for 10 songs, and with Whisper on your own machine the whole video costs nothing. 1 credit = $0.01. Checked against the public price list on 2026-11-11.

Paste this into Claude Code, Codex or Grok:

Make a lyric video for ./song.mp3 with the lyrics in ./lyrics.txt: 9:16, karaoke style,
the sung line big and each word lit as it is sung.
Follow https://varg.ai/how-to/make-a-lyric-video.md.
Show me a timing proof and six frames before the full render.

The agent needs two skills, installed once (HyperFrames renders the video, varg gets the word timings):

curl -fsSL https://varg.ai/install | sh && vargai install hyperframes varg-ai

What goes in, what comes out

InOut
The song: an MP3 or WAV you own or have licensedOne MP4 the length of the song, 1080×1920 (9:16) or 1920×1080 (16:9)
The lyrics as textEach line on screen when it is sung, the current word lit, karaoke style
Optional: cover art or footage for the background, and a lookThe project folder, so a wrong line is a text edit and a new render

How it works

  1. Word timings from the vocal. One varg call (the transcription tool, whisper_large_v3, your lyrics as the prompt) returns a start and an end for every word: 5 credits. Without a key, npx hyperframes transcribe song.mp3 does it on your machine (add --model small --language es for a song that is not in English).
  2. Lyrics drawn in code. HyperFrames renders HTML to video, so the type is real text: npx hyperframes init lyric-video --resolution=portrait --audio song.mp3 --skip-transcribe --non-interactive --skip-skills (--resolution=landscape for 16:9), then the karaoke block, npx hyperframes add caption-pill-karaoke.
  3. A look before the render. npx hyperframes check && npx hyperframes snapshot --frames 6: the agent reads the frames and fixes a word that shows before it is sung or a line that runs off the screen.
  4. Render. npx hyperframes render --output lyrics.mp4, on your machine, as many times as it takes.

Every command and the exact API call: How to make a lyric video with AI.

Looks

These are presets of the motion-formats skill (npx skills add vargHQ/skills --skill motion-formats). They render locally and cost nothing.

What it is not

Music and lyric videos for your song with AI agents

For artists, producers and labels: an AI music video directed shot by shot to your song, a lyric video, a beat-cut edit or a photo slideshow, with lip-sync on every sung line, made by Claude Code, Codex or Grok.

Start with the overview: How to make an AI music video

Make

Edit