Audio Waveform Generator

Make a waveform image from any audio file

The Audio Waveform Generator draws the shape of a song, podcast or voice note as a clean graphic. Choose bars, a mirrored “reflection” look, a flowing line or the classic filled waveform; set your colours or a gradient, pick a size for YouTube, Instagram, a podcast cover or a website header, and download a PNG (with a transparent background if you like), a scalable SVG or a JSON file of peaks for developers.

Podcasters use waveform images on episode artwork and audiograms, musicians on single covers and merch, designers in posters and wedding prints of a first-dance song, and developers to show a waveform in a web player without decoding the audio in the browser each time. You can also turn a clip of up to a minute into a short MP4 with a moving playhead, ready to share on social media.

The audio is read on your device and never uploaded, so it works with unreleased tracks and private recordings.

How to use the Audio Waveform Generator

1. Drop an audio file (MP3, WAV, M4A, OGG, Opus or FLAC) onto the box, or click to choose one.
2. Pick a style — Bars, Mirrored, Line or Filled — and a size preset or a custom width and height.
3. Set the wave colour, a gradient end colour and the background, or tick Transparent background.
4. Fine-tune bar width, gap, padding and wave height; type a Start and End time to draw only part of the file.
5. Press Download PNG or Download SVG, export JSON peaks, or Make MP4 video for a clip.

Audio Waveform Generator features

Which style and settings to choose

• Bars read clearly at small sizes — try 6–10 px bars with a 3–5 px gap for social posts, or 2–3 px bars with no gap for a detailed strip.
• Mirrored puts the bars on a baseline with a faded reflection underneath, a popular look for music players and covers.
• Line draws a single smooth stroke that swings up and down with the loudness — good for minimalist artwork and logos.
• Filled is the precise waveform an audio editor shows, using the real positive and negative peaks of the signal.
• RMS follows perceived loudness and hides single sharp peaks; Peak shows the true maximum level. Heavily compressed modern masters look like a solid block in either mode — reduce the wave height for breathing room.

How the waveform is calculated

The file is decoded in the browser with Mediabunny (or the Web Audio API) and reduced, in a single pass, to the minimum, maximum and sum of squares of every short block of samples — up to 2,000 blocks per second — for the left channel, the right channel and their mix. Each bar, point or pixel column then combines the blocks in its slice of time: the peak is the largest absolute sample, and the RMS value is the square root of the mean square, multiplied by √2 so a steady sine wave reaches the same height in both modes.

The audiowaveform JSON stores each point as a minimum and a maximum scaled to −128…127, with samples_per_pixel telling a player how much audio each point covers, so a waveform can be shown before the audio itself has loaded.

Tips for great-looking waveforms

• Trim silence and long intros first with the Audio Cutter, or use Start and End to draw only the chorus.
• Match the gradient to your artwork, and leave 6–10% padding so the shape does not touch the edges.
• For a print, export SVG and place it in your layout so it stays crisp at any size.
• Want to see pitch rather than loudness? Open the same file in the Audio Spectrogram.