Video & Audio
Audio Spectrum Analyzer & Spectrogram
See sound frequencies live from your microphone — real-time spectrum bars and a scrolling spectrogram, with the loudest frequency and note shown.
How to use
- 1Press Start and allow microphone access.
- 2Make a sound — the spectrum and spectrogram update live. Hover to read frequencies.
Frequently asked questions
What is a spectrogram?
A picture of sound over time: frequency on the vertical axis, time horizontally, and color for loudness.
What can I use it for?
Finding hum or feedback frequencies, checking whistles or tones, teaching physics, or just seeing your voice.
Is the audio recorded?
No — it's analyzed live in your browser and never stored.
Why the frequency axis is logarithmic, and what that does to the display
The spectrum bars map frequency to horizontal position using log10(f/20)/3, which specifically sends 20Hz to the far left (position 0) and 20,000Hz to the far right (position 1, since log10(20000/20) = log10(1000) = 3) — a logarithmic scale rather than a linear one. This matters because human hearing itself perceives pitch logarithmically: the octave from 100Hz to 200Hz sounds like the same-sized musical jump as 1000Hz to 2000Hz, even though the second gap is ten times larger in raw Hz. A linear frequency axis would visually compress all the musically important low and mid frequencies into a tiny sliver on the left while wasting most of the display width on the high end — the log scale instead gives proportionally equal visual space to each octave.
A worked example of how loudness becomes bar height and spectrogram color
The analyser reports each frequency bin's level in decibels, typically ranging from about -110dB (near silence) to -10dB (quite loud) for a typical microphone signal — the display maps that range to a 0-to-1 bar height with (value + 110) / 100, so a bin reading -60dB becomes (−60+110)/100 = 0.5, a half-height bar. The spectrogram uses a similar but separately-tuned mapping to pick a color (shifting through a hue range from blue through red as loudness increases) for each pixel in its newest column, which is redrawn every frame while the previous columns are shifted two pixels to the left — building the scrolling waterfall effect where you can watch a sustained tone leave a solid horizontal streak and read its pitch directly from its vertical position.
Why this can't just use Nyquist and show 0-22kHz as it comes
The actual frequency resolution available depends on the audio hardware's sample rate and the FFT size the analyser is configured with — the code explicitly stops drawing content past 20,000Hz regardless of what the underlying hardware could technically report, because 20kHz is the conventional upper edge of human hearing and displaying content above it would just be wasted space showing inaudible ultrasonic content most microphones can't even capture cleanly anyway. The practical range that matters for voice, music and most real-world sound analysis is comfortably covered by the 20Hz-20kHz window this tool deliberately commits to.
Echo cancellation and automatic volume are left off, so the hum stays visible
The microphone is opened with echo cancellation, noise reduction, and automatic volume all turned off. Those browser features would otherwise notch out a hum, ride the loudness, and soften the peaks this display is here to show. The picture is smoothed as well: each update keeps about 60% of the previous frame and mixes in 40% of the new reading, so the bars do not flicker on every syllable. On a typical 48,000 Hz microphone, each bar covers about 5.9 Hz, and nothing is drawn above 20,000 Hz. Freeze does not turn the microphone off. Both pictures hold still, and the microphone stays open until you press Stop.
The loudest note skips rumble at 40 Hz and anything quieter than −80 dB
The Loudest line is not simply the tallest bar at the left edge. It looks only above 40 Hz, keeps the loudest frequency it finds, and appears only when that level is louder than −80 dB. Quieter than that, the line stays hidden. A strong 440 Hz tone is reported as the note A4. On the spectrogram, the top of the picture is 20,000 Hz and the bottom is 20 Hz. Color fills in sooner than the bars do. A tone at −20 dB already uses the strongest spectrogram color, while that same tone reaches only about 90% of the bar height. Each moment, the spectrogram shifts a little to the left and the newest slice appears on the right.
Related tools
- Sound Level Meter (Decibel Meter)
Measure noise levels with your microphone — live decibel reading, peak, average and a history graph — with calibration and noise-exposure guidance.
- Audio Speed Changer (Keep Pitch)
Speed up or slow down audio from 0.5× to 2× — keep the original pitch (time-stretch) or change it like a tape — for podcasts, lessons and practice.
- Audiogram / Waveform Video Maker
Turn audio into a shareable video with an animated waveform, title and cover image — perfect for promoting podcast clips on social media.
- Mute Video (Remove Audio)
Remove the sound from a video instantly — no re-encoding, no quality loss — and download the silent MP4.
- Online Screen Recorder
Record your screen, a window or a browser tab — with optional microphone — right in your browser, then download the video. No install, no watermark.
- Online Voice Recorder
Record your voice or any microphone in the browser — pause, play back and download the recording. No sign-up and nothing is uploaded.