Audio Visualizer: Waveform & Spectrogram PNG from Any Audio
Turn a recording into a shareable PNG: an editor-style waveform, a log-frequency spectrogram, or both stacked on one time axis.
Sound is invisible, but its structure is not: a waveform shows when things happen and how loud they are, and a spectrogram shows which frequencies carry the energy from moment to moment. This tool decodes a recording and draws both the way editors do: the waveform as a light peak envelope with a darker loudness core, the spectrogram as Hann-windowed frequency analysis on a log axis with 100 Hz, 1 kHz and 10 kHz marked, brighter always meaning louder. Choose either view or the stacked pair on one shared time ruler, pick a frequency-detail level and an image width, and download the result as a PNG for a thumbnail, a podcast episode card, a bug report about a hum, or a lesson on what a chord looks like.
How to Use It
- Upload your audio file and wait for the local preview to load.
- Adjust the settings: Choose the view, waveform, spectrogram or both stacked. Then set the frequency-detail level, the theme and the image width; the height follows from the view.
- Download the result: Download the processed file directly to your device.
Controls and Options
Pick the stacked view to read events across both pictures at once: a kick drum shows as a spike in the waveform and a bright vertical stripe in the spectrogram at the same position on the shared time ruler. Raise the frequency detail to 4096 to separate close pitches such as hum against a bass line, or drop it to 1024 when the timing of clicks and consonants matters more than pitch separation.
Formats and Output
MP3, WAV, M4A, OGG, and FLAC inputs are decoded in the browser. The result exports as a PNG image at 1280, 1920 or 2560 pixels wide, with the height set by the chosen view.
Frequently Asked Questions
What is the difference between the waveform and the spectrogram?
The waveform plots level against time: the lighter outline is the true peak envelope and the darker core is the RMS level, which tracks how loud the audio actually sounds. The spectrogram plots frequency against time, with brightness showing how much energy each frequency carries: a melody appears as a moving line, a chord as a stack of lines, a click as a thin vertical stripe.
How is the spectrogram computed?
The audio is cut into overlapping Hann-windowed slices of the chosen analysis size, each slice goes through a Fourier transform, and the magnitudes are converted to decibels with a -90 dB display floor. Slices overlap by half their length, so every part of the recording is weighted evenly. Columns are then reduced to the image width by keeping the loudest value, so short transients stay visible instead of being averaged away.
Why is the frequency axis logarithmic?
Hearing works in ratios: the step from 100 to 200 Hz sounds as large as the step from 1 to 2 kHz. A log axis gives each octave the same height, which is why the 100 Hz, 1 kHz and 10 kHz gridlines are evenly spaced and why melodies and chords keep their visual shape wherever they sit in the range.
Which frequency-detail setting should I pick?
The window size trades time detail against frequency detail. 1024 keeps drum hits and consonants crisp but smears bass notes together; 4096 separates close pitches, two singers a semitone apart, mains hum against a bass line, but blurs fast events; 2048 is a sensible middle for most music and speech at common sample rates.
How large is the exported image?
You choose the width, 1280, 1920 or 2560 pixels, and the height follows from the view: the waveform strip is width x 9/32, the spectrogram is half the width, and the stacked pair is width x 5/8, so 1920 wide gives 540, 960 or 1200 pixels tall respectively. The result is a standard PNG that any editor, chat app or document accepts.
Is my audio uploaded anywhere?
Supported audio content is not uploaded for processing. Decoding, frequency analysis and image rendering run locally in your browser; site analytics and the assistant are separate.