Vizard Agent

How to make a lyric video with beat-synced screen shaking

Last updated 2026-08-18 · 6 min read

Upload your track to Vizard Agent and say you want the screen to shake harder on the heavy parts. Vizard Agent transcribes the vocals with word timings, measures the song's energy and harshness across its whole length, and ties the shake to those measurements so it intensifies where the track actually gets heavier.

What is the short version?

A lyric video is a solved problem right up until you want the picture to react to the music. "Shake more on the growls" is easy to say and impossible to do by hand without listening to the whole song with a notepad, twice.

  1. Go to Vizard Agent and upload the track and the artwork.
  2. Say the effect you want and where it should get stronger.
  3. Say the shape and whether the lyrics should be styled to the genre.

What do you need before you start?

The track and a description of the reaction. Vizard Agent finds the words and measures the intensity itself, so what it needs from you is the rule connecting them: which parts of the song should look different, and how much different.

What do you type into Vizard Agent?

Describe the rule, not the timestamps. Vizard Agent measures the song and works out where the rule applies, so "more shaking when it gets heavy" is a complete instruction and a list of timecodes is work you do not need to do.

Prompt

Make a lyric video from this track. Add a slight shake throughout, and make it shake more where there is [growling or heavy sections].

Variants worth knowing:

What does Vizard Agent actually do?

Here is the order Vizard Agent worked in on a real lyric video. The step that makes the whole thing work is a measurement, not an effect: it builds an intensity curve for the song and drives everything from that.

  1. Probes the uploaded file and checks the audio track.
  2. Grabs sample frames and a full-resolution frame of the artwork.
  3. Loads the music-video playbook rather than improvising a structure.
  4. Transcribes the vocals with word timings.
  5. Analyses the song structure and locates the growl sections.
  6. Reads the full transcript and checks the caption styling options.
  7. Measures the song's energy and harshness envelopes across its whole length.
  8. Browses the available fonts and reads every word-level timing.
  9. Crops the artwork for the background and the logo, indexes every sung word, and builds the lyric line files against the measured curves.

Step 7 is the whole technique. "Where it gets heavy" is not a section boundary you can guess from the structure; it is a measurable property of the audio, and once it is measured the shake can follow it continuously rather than switching on and off at a marker.

What does the result look like?

From the run this page is written from, probed on the delivered file: 1920x1080, H.264, 24fps, 267.01 seconds, AAC audio. Wide and four and a half minutes, the full song, with lyrics timed to the words and a shake that tracks the track's own intensity.

Four minutes is unusual for anything else on this site. A lyric video is one of the few formats where full length is the point, since it is made to be listened to rather than scrolled past.

Four and a half minutes is the whole song. Vizard Agent times every line against the vocal rather than against a fixed grid, which is why the lyrics stay locked to the words through tempo changes and instrumental passages.

When does this not work well?

The whole effect depends on the audio being measurable and the lyrics being transcribable, and heavy music is exactly where both of those get harder. Vizard Agent measures the intensity rather than guessing at it, and some vocals defeat transcription entirely no matter what.

How do you fix a result that came back wrong?

Say which part and which direction. Vizard Agent keeps the transcript, the word timings, the measured envelopes and the lyric line files, so dialling the effect down or restyling one section works against curves it has already computed.

How does Vizard Agent compare to doing it yourself?

By hand this is transcribing the lyrics, timing every single line, then listening through four and a half minutes with a notepad marking every point where the track gets heavier, and finally keyframing an effect against all of those marks by hand.

By hand Vizard Agent
Lyric timing Line by line Word-level timings from the vocal
Finding the heavy parts Listen with a notepad Energy and harshness measured
The effect Keyframe each section Driven by the measured curve
Toning it down Re-keyframe everything Same curve, lower amplitude

Common questions

Can it transcribe screamed vocals? Poorly. Paste the lyrics for those sections and Vizard Agent times them against the audio.

Does it work for the whole song? Yes. This is one of the few formats where full length is normal rather than excessive.

Can the font change with the music? Yes, and it is effective. Bigger type on the heavy parts is tied to the same measurement.

What if the shake is too strong? Say so. Vizard Agent lowers the amplitude against the same curve rather than rebuilding.

Can I use my own artwork? Yes. Upload it and Vizard Agent crops it for the background and pulls the logo out of it as a separate element.

Can it style the lyrics to the genre? Yes. Vizard Agent browses the installed fonts and picks one that suits the track rather than defaulting.

Who clears the music? You do. Vizard Agent builds the video around whatever track you upload and does not check the rights.

Can it handle tempo changes? Yes. The timings come from the vocal itself rather than from a fixed grid.