# Waveform Image from Audio Turn any audio file into a waveform PNG in your browser — pick the size, solid or gradient wave colors and background, or keep it transparent. Nothing is uploaded. ## Run it - **CLI:** `gizza tool waveform-image 'url=https://example.com/input' 'width=1200' 'height=300' 'color=#4f46e5' 'split_channels=true' 'scale=lin' 'sampling=average'` - **Web:** https://gizza.ai/tools/waveform-image/ ## Inputs - `audio` — Audio _(file; accept: audio/*)_ - `width` — Width (px) _(field)_ - `height` — Height (px) _(field)_ - `color` — Wave color _(field)_ - `color2` — Gradient end (optional) _(field)_ - `background` — Background _(field)_ - `split_channels` — One lane per channel _(field)_ - `scale` — Amplitude scale _(field)_ - `sampling` — Sampling _(field)_ ## Output - Waveform image (image) ## Query parameters Open the tool pre-filled and auto-run via URL: - `width` — Width (px) - `height` — Height (px) - `color` — Wave color - `color2` — Gradient end (optional) - `background` — Background - `split_channels` — One lane per channel - `scale` — Amplitude scale - `sampling` — Sampling - `url` — fetch the input file from a public URL (CORS-permitting) Example: `https://gizza.ai/tools/waveform-image/?url=https://example.com/input&width=1200&height=300&color=%234f46e5&split_channels=true&scale=lin&sampling=average` --- ## Make a waveform image from any audio file Upload an audio file and this tool renders its waveform as a PNG image — the classic soundwave graphic used for podcast covers, social banners, audio-post thumbnails and player artwork. Choose the image size, a solid or gradient wave color and an optional background color, or leave the background empty for a transparent PNG that drops onto any design. The render runs entirely in your browser with ffmpeg compiled to WebAssembly (its `showwavespic` filter), so your audio is never uploaded to a server. ### Worked example Make a social banner for a podcast episode: upload the episode mp3, leave **Width** and **Height** empty (the default is a banner-shaped `1200×300`), set **Wave color** to `#4f46e5` and **Background** to `#0b1220`. The result is a 1200×300 PNG named like the original with a `-waveform.png` suffix — dark card, indigo wave. Add `#ec4899` as the **Gradient end** and the wave fades indigo→pink left to right. For a transparent overlay to place on top of your own artwork, leave **Background** empty instead: the PNG keeps its alpha channel, so only the wave itself is opaque. A quiet voice memo that renders as a thin line becomes clearly visible with **Amplitude scale** set to `sqrt` or `log` — and **Sampling** set to `peak` keeps short hits visible. Or just tap one of the **Try** presets under the fields. ### Options - **Width / Height** — image size in pixels (16–4096 × 16–2048). Empty fields use the 1200×300 default. - **Wave color** — hex color: `#RGB`, `#RRGGBB` or `#RRGGBBAA` (the last two digits are alpha, for a translucent wave). Pick it with the swatch or type it. With **One lane per channel**, a comma-separated list like `#4f46e5,#ec4899` colors each channel's lane (up to 8). - **Gradient end** — a second hex color. When set, the wave is filled with a horizontal left→right gradient from **Wave color** to this color. Leave empty for a solid wave. - **Background** — hex color, or empty for a transparent PNG. An alpha hex like `#00000080` bakes in a see-through scrim. - **One lane per channel** — off (default) downmixes to a single clean mono wave; on draws each channel (e.g. stereo left/right) in its own horizontal lane, sharing the image height. - **Amplitude scale** — `lin` is the true waveform; `sqrt`, `cbrt` and `log` progressively boost quiet material so it stays visible. - **Sampling** — `average` (default) draws the mean level of each pixel column; `peak` draws the loudest sample instead, a fuller wave that keeps drum hits and other transients visible. ### Limits and edge cases - Input files up to 10 MiB; anything ffmpeg can decode works (mp3, wav, flac, m4a/aac, ogg, opus — and most video containers' audio tracks via the CLI). - The output is always a PNG (with an alpha channel when the background is empty). Colors must be hex — named colors like `red` are rejected with a hint rather than guessed. Short `#f00`-style hex is expanded for you. - A gradient (**Gradient end** set) needs a single **Wave color**, not a comma list — the gradient runs across the whole image, so split-channel lanes all share it. - Very quiet recordings can look like a flat line in `lin` scale — that's the honest amplitude, not a bug; switch the scale to `sqrt` or `log`, or set **Sampling** to `peak`. - A silent file renders only the thin center baseline. - One image per run: the whole track is drawn into the width you choose, one peak column per horizontal pixel — longer audio isn't wider, it's just more compressed. ## FAQ
How do I get a transparent background? Leave the **Background** field empty (the default). The PNG then keeps its alpha channel, so everything except the wave itself is fully transparent — ideal for dropping the waveform onto your own artwork or a colored card in any editor. Set a hex value like `#0b1220` only when you want a flat background baked into the image, or an alpha hex like `#00000080` for a half-transparent scrim.
Can the wave be a gradient? Yes — set **Gradient end** to a second hex color. The wave is then filled with a smooth horizontal gradient that starts at **Wave color** on the left and ends at **Gradient end** on the right (try `#f97316` → `#ec4899` for a sunset look). Everything else — transparent or solid backgrounds, channel lanes, scales — works the same as with a solid color.
Why does my quiet recording show almost no waveform? The default `lin` scale draws true amplitude, and quiet audio genuinely has small peaks. Switch **Amplitude scale** to `sqrt` — or `log` for the strongest boost — to make quiet material fill more of the image height without touching the audio itself. Setting **Sampling** to `peak` helps too: it draws each column at the loudest sample instead of the average, so short transients stop disappearing.
What size should I pick for social media or a podcast player? The default `1200×300` works well for wide banners, link previews and audio-post headers. For a square post, try `1080×1080`; for a compact player strip, something like `800×160`. Any size from 16×16 up to 4096×2048 works — the whole track is always drawn across the full width. The **Try** presets under the fields cover the common shapes.
Can I show the left and right channels separately? Yes — tick **One lane per channel**. A stereo file then renders two stacked waves (left on top, right below) sharing the image height, and a comma-separated **Wave color** list like `#4f46e5,#ec4899` gives each lane its own color. Off, the channels are downmixed to one mono wave, which is usually the cleaner look for artwork.
Can I make bar-style or circular waveforms? Not with this tool — it draws the classic continuous waveform (ffmpeg's `showwavespic`), in solid or gradient color. Bar, dotted and radial "sound-wave art" styles need a different renderer and aren't supported here. What you can vary: size, colors (including alpha), gradient, background, per-channel lanes, amplitude scale and peak sampling.
Is my audio uploaded anywhere? No. The page downloads an ffmpeg WebAssembly build once and then renders the image locally in the browser tab — the audio never leaves your device.