how do I make a music visualizer?

In kott: pick a visual source (a still image such as album art, a video file, or a live camera), load your track as an audio file, choose an effect that reads the audio — spectrum-bars for the classic FFT bar graph, oscilloscope for the waveform — then add a modulation node so a signal you choose moves a parameter you choose. Press REC and play the track through. REC writes the canvas to an MP4 with the audio in sync, so the recording is the take you performed. There is no upload-and-render step: the visuals are drawn live and you capture them.

is there a free music visualizer?

In kott, every effect previews free with no account, and a free set exports without paying — but the audio-reactive effects, spectrum-bars included, need the one-time licence. So you can open the studio, load a track and watch the bars move for nothing, and paying is what lets you record it. If you need free and fully automatic, what you want is an audio-to-video renderer that takes an upload and returns a file, which is a different kind of tool from this one.

can I use my own song?

Yes. kott takes an audio file you load from your own disk, and it also takes a live microphone or system audio if the track is playing out of a DAW or another application. Everything is processed on the device — the audio is decoded and analysed in your own browser and nothing uploads.

can I make a music visualizer in a browser?

Yes. kott runs in any browser with WebGL2, with nothing to install and no account needed to look. The spectrum analysis, the effects and the recording all happen on your machine. There is also a native macOS app that runs fully offline, covered by the same one-time licence as the web studio.

does it export video?

Yes. REC writes the canvas to an MP4 with the audio in sync, recorded in real time as the track plays. Stills come out as PNG, and vector paths are available for print and plotters. There is no batch export and no render queue — one take at a time, at the length of the material.

what makes it different from a template visualizer?

The reaction is not decided for you. kott derives spectral-flux onset, a phase-locked beat pulse, a downbeat, build and drop detection, tempo between 60 and 180 BPM, bass/mid/treble and 2 to 8 EQ bands you define yourself, and any of those signals can be routed onto any parameter of any effect through a modulation node with its own gain and smoothing. You choose what the bass moves. That is also the cost: nothing looks finished until you have wired something.

music visualizer, played live

Read this before anything else: kott does nottake an MP3 and hand you back a rendered video. There is no upload step, no render queue, no template gallery, no lyric captions and no batch processing. kott draws the visuals live from the audio and you record the performance. If what you want is hands-off — give it a file, come back to a finished clip — kott is the wrong tool, and an automated audio-to-video renderer is the right kind of thing to look for.

What it is instead: load an audio file, or use the microphone or system audio. Pick a source image — album art works — or a video clip or a live camera. Choose spectrum-bars for the classic FFT bar graph, or oscilloscope for the waveform. Route the bass onto a parameter so the kick moves the thing you chose it to move. Then press REC, which writes the canvas to an MP4 with the audio in sync.

The second thing to know early: the audio-reactive effects on this page are paid. Every effect previews free with no account and a free set exports without paying, but spectrum-bars and the rest of the spectacle need the licence — $39 until 13 September, once, for the web studio and the macOS app together, with no subscription. Everything runs in the browser on your own machine, and nothing uploads.

open spectrum bars →

01/ INSTEAD

when to leave

when kott is the wrong tool for this

“Music visualizer” covers more than one job, and most tools in the results do the other one. Here is when yours is the other one:

You want to upload an MP3 and get a video back. kott has no render queue and no upload step. The audio is played on your machine and the visuals are drawn live from it; a file exists only once you have recorded one.
You want a template. There is no template gallery, no preset scene to drop a track into, and no starting point that already looks finished. You pick a source image or clip and build the look yourself.
You want lyric captions. kott does not read lyrics, does not sync text to a vocal and has no caption track. Anything typed on screen is typed by you.
You want a batch. There is no queue and no way to point kott at ten tracks and come back later. One track, one performance, one recording.
You want it free. Every effect previews free with no account, and a free set exports without paying — but the audio-reactive spectacle, spectrum-bars included, needs the one-time licence. If free-and-automatic is the requirement, look for an automated audio-to-video renderer instead. That is a different kind of tool and it is a reasonable thing to want.

02/ MAKING

step by step

what making one actually looks like

It happens in one browser tab, in this order.

give it something to look at: The visual source is a still image, a video file, or a live camera. Album art is the obvious still, and it is a good one — spectrum-bars tints its bars with the source image, so the cover's own colours end up in the graph rather than a stock gradient. A short clip works the same way and loops under the audio.
load the track: Drop an audio file in and it plays on a loop while you work. The other two inputs are the microphone and system audio, which is what you use when the track is coming out of a DAW or a browser tab rather than off disk.
pick the thing that reads the audio: spectrum-bars is the classic: a vertical FFT bar graph where x is frequency and each bar's height is that band's magnitude, with a soft peak cap riding the top. oscilloscope draws the time-domain waveform instead, as a glowing horizontal trace with a gaussian bloom. Both respond to the track immediately, before you have set anything.
route one signal onto one parameter: This is the step that stops it looking like everyone else's. Add a modulation node, choose a bass band as the source, choose a parameter as the destination — the bar scale, the glow, the warp amount — and set the node's gain and smoothing. Now the kick moves that one control and nothing else. Do it twice more and the piece has a structure.
hit REC and play the track through: REC writes the canvas to an MP4 with the audio in sync. What you recorded is what you did while it was running, including anything you moved by hand, so a take is a performance rather than a render. Stills come out as PNG, and vector paths are available for print and plotters.

03/ VISUALS

named, with uses

the visuals that read the audio directly

These take the spectrum or the waveform as their input, so they move the moment a track is playing and before you have set anything.

spectrum-bars: A vertical FFT bar graph. The horizontal axis maps to frequency and each bar's height comes from the spectrum at that frequency, so the low end sits on the left and the air on the right. The source image tints the bars, and a soft peak-cap glow rides the top of each one so the bar tops do not read as a flat cut-out. This is the effect most people mean when they say music visualizer.
oscilloscope: The time-domain waveform, drawn as a glowing horizontal trace with a gaussian bloom around it. Where spectrum-bars shows you what frequencies are present, this shows you the actual shape of the wave, which is why it reads as tighter and more legible on sparse material — a solo voice, a synth line, a kick with space around it.
radial-spectrum and freq-rings: The same spectrum data wrapped around a centre instead of laid out on a line. radial-spectrum spokes the bands out from the middle; freq-rings puts them in concentric rings. Both suit a square crop and album art better than a horizontal bar graph does.
reactive-warp and signal-wave-plus: Rather than drawing the audio as a graph, these push the source image around with it — the picture bends and ripples on the signal instead of a chart appearing over it. This is the route to something that does not announce itself as a visualizer.
datamosh-real: Compression artefacts driven by the audio: blocks smear and the frame tears where the signal spikes. It is the loudest option here and the one that most needs a parameter routed by hand rather than left at its default.

04/ ROUTING

you choose

you choose what the bass moves

Every tool in this space says “audio-reactive”. The question worth asking is which signal reaches which control, and whether you get to decide.

what kott hears, beyond the raw spectrum: Spectral-flux onset with a phase-locked beat pulse. A downbeat on the first beat of every four-beat bar. Build and drop detection. Tempo tracked between 60 and 180 BPM. Bass, mid and treble as separate signals, plus 2 to 8 EQ bands you define yourself, each with its own frequency window, gain and gate.
any of those lands on any parameter: kott does not have a list of audio-reactive effects with the reaction baked in. It has modulation nodes: one picks a signal as its source and an effect parameter as its destination, and carries its own gain and smoothing. Every parameter of every effect is a valid destination, not only the ones on this page.
which is why it does not look like a preset: A template decides for you that the bars move with the volume. Here you decide that the second EQ band moves the bar scale, the downbeat moves the palette, and a build moves the warp — or none of that, and something else entirely. Two people using the same effect on the same track produce different pieces, because the wiring is the work.
or put a parameter under a physical fader: WebMIDI, so a controller works in the browser with nothing installed. A nanoKONTROL2 map ships with it and any control MIDI-learns onto any parameter. Recording a take with one hand on a fader is a different result from recording one that runs itself.

05/ LIMITS

and where to go

what it will not do

No upload-and-render. There is no queue, no server-side rendering and no finished file waiting for you. You record the performance, in real time, at the length of the track.
No templates and no lyric captions. Nothing arrives pre-composed, and there is no text track tied to the vocal.
The audio-reactive effects are paid. Previewing every effect is free and needs no account, and a free set exports without paying, but spectrum-bars and the rest of the spectacle need the one-time licence.
No NDI, no Syphon, no virtual camera. kott cannot feed video into another application, appear in OBS as a source, or present itself as a camera to a meeting.
The visual language is narrow on purpose: dither, halftone, print, ASCII, engraving, glitch, CRT. No 3D, no generative geometry, no particle scenes built out of nothing. If you want a rendered 3D world reacting to a track, this is the wrong instrument.
It is not a video editor. No timeline, no clips, no cuts. A recording comes out as one continuous take.
The offline native build is macOS only. On Windows and Linux kott means a browser tab with WebGL2, which is enough for all of the above, but there is no app to install.

nearby pages, and what each one answers

the routing itself, for performing: audio reactive visuals — the modulation graph treated as an instrument: signals, destinations, smoothing and MIDI, written for someone playing to a room rather than producing a clip for one track.
choosing a tool at all: vj software — the category page, with real prices for the field and a “buy the other one when” list against each entry.
a camera instead of an image: webcam dither — the same machine with a live camera as the visual source, dithered in real time on the GPU, and an honest account of why a moving image is harder than a still.

06/ FAQ

asked as typed

questions

how do I make a music visualizer?: In kott: pick a visual source (a still image such as album art, a video file, or a live camera), load your track as an audio file, choose an effect that reads the audio — spectrum-bars for the classic FFT bar graph, oscilloscope for the waveform — then add a modulation node so a signal you choose moves a parameter you choose. Press REC and play the track through. REC writes the canvas to an MP4 with the audio in sync, so the recording is the take you performed. There is no upload-and-render step: the visuals are drawn live and you capture them.
is there a free music visualizer?: In kott, every effect previews free with no account, and a free set exports without paying — but the audio-reactive effects, spectrum-bars included, need the one-time licence. So you can open the studio, load a track and watch the bars move for nothing, and paying is what lets you record it. If you need free and fully automatic, what you want is an audio-to-video renderer that takes an upload and returns a file, which is a different kind of tool from this one.
can I use my own song?: Yes. kott takes an audio file you load from your own disk, and it also takes a live microphone or system audio if the track is playing out of a DAW or another application. Everything is processed on the device — the audio is decoded and analysed in your own browser and nothing uploads.
can I make a music visualizer in a browser?: Yes. kott runs in any browser with WebGL2, with nothing to install and no account needed to look. The spectrum analysis, the effects and the recording all happen on your machine. There is also a native macOS app that runs fully offline, covered by the same one-time licence as the web studio.
does it export video?: Yes. REC writes the canvas to an MP4 with the audio in sync, recorded in real time as the track plays. Stills come out as PNG, and vector paths are available for print and plotters. There is no batch export and no render queue — one take at a time, at the length of the material.
what makes it different from a template visualizer?: The reaction is not decided for you. kott derives spectral-flux onset, a phase-locked beat pulse, a downbeat, build and drop detection, tempo between 60 and 180 BPM, bass/mid/treble and 2 to 8 EQ bands you define yourself, and any of those signals can be routed onto any parameter of any effect through a modulation node with its own gain and smoothing. You choose what the bass moves. That is also the cost: nothing looks finished until you have wired something.

07/ OUT

load a track

put a track through it

The studio opens on spectrum-bars. Drop in an audio file, drop in a cover image, and the bars are already moving with the track. No account, no install, and the preview costs nothing — the fastest way to find out whether a live visualizer is what you wanted is to watch one respond to your own song.

open spectrum bars →