Does my audio get uploaded anywhere?
No, and that is not a marketing claim, it is how the tool is built. The sound wave tool on ilove3d decodes your file with the browser's own Web Audio API, on an offline context, and in the same step reduces it to an envelope of 4,000 energy values. That summary, how loud the clip is every 45 milliseconds, is the only thing that survives: the decoded audio is never stored, never copied, and never sent anywhere, because there is no server to send it to. You can disconnect from the internet once the page has loaded and keep generating and downloading the STL. It is the rule behind every tool on the site: your files stay on your machine. For a voice note, a message from someone who is gone, or a copyrighted song clip, that matters more than convenience.
How long a clip should I pick?
Short: five to fifteen seconds. This is the setting that decides whether the piece is recognizable or a flat band, and the reason is arithmetic. With 60 bars over 15 seconds each bar averages 250 milliseconds, roughly one syllable, so the rhythm of speech survives: the pauses, the emphasis, the laugh at the end. Spread those same 60 bars over a three-and-a-half-minute song and each bar summarizes 3.5 seconds, all of them landing on nearly identical average energy, which prints as a fuzzy rectangle with no shape. That is why the sound wave tool on ilove3d draws the whole file in a scrubber with a window you drag with the mouse or nudge with arrow keys, plus a button to hear the selection before building anything. The start detector finds the first signal above −45 dBFS and leaves 0.15 seconds of air so the opening consonant is not clipped.
How do I print a sound wave plaque?
Flat on the bed, with no supports at all. The audio amplitude is mapped to the length of each bar, inside the plane of the plate, never to its height: everything grows upward from a flat surface, so there is nothing to bridge and nothing to overhang. PLA at 0.2 mm layers and 15% infill is plenty; a 150 × 50 × 4 mm plaque takes about an hour. Thicknesses are quantized to multiples of 0.2 mm on purpose, so the boundary between plate and relief lands exactly on a layer change: that is where you insert the filament swap, with M600 or your slicer's pause button, and the wave comes out in one color over a background in another, which is how the ready-made version looks. Raised text falls in the same swap when you keep it at that height.
Which audio formats does it accept, and how big can the file be?
Whatever the browser can decode: mp3, wav, m4a, mp4, aac, flac, ogg, opus, webm and aiff, up to 30 MB per file. That ceiling is not the sound wave tool on ilove3d being cautious, it is the Web Audio API having no partial decode: even if you only want 15 seconds, the browser decodes the entire file into memory, and a five-minute mp3 at 320 kbps turns into well over a hundred megabytes of samples, which kills the tab on a phone. Files like wma, amr and midi are rejected before anything is attempted, since no browser opens them. Separately, Safari cannot decode ogg or opus even though it plays them, so on iPhone or Mac convert those to mp3 or m4a first. A WhatsApp voice note exported as m4a, or a memo recorded on your phone, loads untouched.
How many bars should I use, and why are they capped?
Sixty bars across 150 mm is the default and a good starting point: it gives a 2.5 mm pitch, with 1.5 mm bars and 1 mm gaps. The cap exists because a 0.4 mm nozzle cannot resolve detail finer than its own extrusion line. Below 1.2 mm of bar width, which is three extrusion lines, the slicer silently drops the bar and you end up with an STL holding fewer bars than the screen showed you; with gaps under 0.8 mm, first-layer squish welds neighboring bars together and the wave prints as one rectangular smudge. So the tool clamps the count to a 2 mm minimum pitch and shows pitch, bar width and gap in millimeters as you move the sliders. In circle style the pitch is measured at the inner radius, where the bars sit closest together.