PULSATRAX Visualizer

Turn your song into an audio-reactive video

Drop in an MP3 or WAV, shape the look live, then export an MP4. Previewing and rendering happen on this device — your song is never uploaded to PULSATRAX.

Free live preview No upload queue Landscape, portrait, square Account needed only for MP4 export
Step 1 of 3 · Choose a song Your MP3 or WAV stays on this device.
Open Studio
Portable project Bundle the song, images, settings, and Scene Director into one local .ptxviz file. No project bundle saved or opened yet.
Local recovery Checking this device for a recovery draft…
No song selected MP3 or WAV · up to 8 minutes
Local only
0:00 / 0:00
Space Play ←→ Seek [] Visual R Surprise P Save PNG F Fullscreen
Song timeline Choose audio to see its waveform.
0:00 Full song 0:00

Look

StyleVisual library
1 / 6
Saved looks and share links never include your audio or image files.
ColorsAccent
Camera & frameComposition
EffectsGlow & depth
SpectrumAudio overlay
Audio responseIntensity

Auto Scene Mix

Build a sequence of proven MilkDrop scenes and snap each change to the strongest nearby beat.

Choose a song, then enable Scene Mix.
Scene director Preview and shape the exact sequence used in your video.

Enable Scene Mix and choose a song to edit its sequence.

No scene changes to undo.

    Background / cover art

    Drop a PNG, JPG, or WebP on this card, or choose one. It stays local and can move with the music.

    No background selected

    Title card

    Artist logo

    Drop a PNG, JPG, or WebP on this card, or choose one. It stays local and can react to the music.

    No logo selected

    Lyrics & captions

    Import timed captions, paste your own words, or search by title and artist. Audio stays local; catalog search shares only those two text fields.
    Catalog match Find the real lyrics Searches LRCLIB with the Song title and Artist from the Title card. Only those two fields are shared — your audio never leaves this browser.
    Enter both song title and artist above.
    Known songs work best with catalog lyrics instead of AI guessing. Lyrics results from LRCLIB. Choose the exact recording before applying.
    Experimental Detect words from the audio For unknown or unreleased songs, separate the singer first and let Whisper listen to that cleaner local layer. Every audio sample stays on this device.
    Vocal focus Separate the singer before detection Optional HTDemucs preparation for dense mixes. First use downloads a 181 MB model; processing may take several minutes and works best with desktop WebGPU.
    Device · checks after song Model · 181 MB on first use
    Choose a song to inspect vocal isolation on this device. Powered by demucs-web, HTDemucs, and ONNX Runtime Web. The model is cached by your browser; no stems are uploaded or stored by PULSATRAX.
    Engine · checks after song Model · cache checks after song
    Choose a song to enable automatic lyrics. Title, artist, and Preferred spellings guide Whisper without being inserted. Close alternatives are corrected transparently after detection — add uncommon names, slang, or words Whisper confuses. Auto · song samples up to 30 seconds to detect language, then uses word timestamps to build short sequential caption phrases. Long songs overlap at quieter cuts, but Automatic lyrics decodes every part independently so one bad or instrumental part cannot poison later minutes. Near-silent parts skip the model, and near-silent cues are discarded. Best uses five-beam Whisper Small (first local download about 300–969 MB); Base is balanced. Automatic lyrics and Sync exact use timestamp models. Failures safely step down to Base, then Tiny. Review unclear words before export.
    Caption timing · click a timed line above to select it Choose or generate timed lyrics first
    No lyrics yet · import a timed file or paste plain lines. Live cueWaiting for lyrics
    MP4 · HD 720p · 30 FPS

    Export locally

    Not checked for these export settings.
    Estimated renderChoose audio
    Resolution1280 × 720
    Frame rate30 FPS
    Estimated size—
    Export storageMP4 is assembled locally in this tab.
    Full song · choose audio first

    Preview everything for free. Create a free account or sign in only when you are ready to export.

    Fast local render — usually much quicker than the song length.

    Choose a song to begin.

    1

    Choose your song

    The browser decodes your MP3 or WAV locally. There is no audio upload request.

    2

    Play with the visuals

    Switch styles, colors, intensity, titles, and social video shape while the song plays.

    3

    Export on your hardware

    After sign-in, WebCodecs renders an MP4 with the best compatible local codecs and downloads it directly.

    Your video setup is saved locally

    Create a free account to export

    You can keep previewing without an account. Sign in or register to unlock MP4 export; this tab will restore the selected song and settings when you return.

    Sign in Create account