See any sound as a live spectrum.
Turn your microphone into a real-time FFT analyzer. Switch between a frequency bar graph, an oscilloscope, and a scrolling spectrogram; read the peak frequency, its musical note, and the level as they change. Everything runs in the browser with the Web Audio API — nothing is recorded or uploaded.
Press Start analyzer to begin. Your browser will ask for microphone permission.
Microphone access is blocked
The analyzer needs mic access to read live audio. Re-enable it, then press Start again:
- Chrome / Edge: click the tune or camera-and-mic icon in the address bar → set Microphone to Allow → reload.
- Firefox: click the mic icon in the address bar and clear the blocked permission, or use Settings → Privacy & Security → Permissions → Microphone.
- Safari: Safari → Settings → Websites → Microphone → set this site to Allow, then reload.
Spectrum analyzer.
Start the mic, pick a view, and watch the frequencies of whatever you play, sing, or say appear in real time.
Check these before you start
- Allow microphone accessThe analyzer reads live audio from your mic, so the browser will ask permission the first time you press Start. If you blocked it before, re-enable it for this site.
- Pick the source you want to seeChoose the right input in the device menu — a USB interface, an instrument mic, or the built-in mic each show a very different spectrum.
- Mind feedback if speakers are loudThe signal is never played back here, but a loud speaker feeding the same mic can still build a howling peak in the room. Use headphones when analyzing playback.
- Start with FFT 2048It's a good balance of frequency detail and responsiveness. Go larger to separate close tones, smaller to catch fast transients — the trade-off is explained below.
How to use the spectrum analyzer
- 1Start the analyzer. Press Start analyzer and allow microphone access. The status flips to Live and the display starts moving with the sound in the room.
- 2Choose a view. Switch between Bars (frequency content, low to high), Scope (the raw waveform in time), and Spectrogram (frequency over time, scrolling) with the toggle or keys 1/2/3.
- 3Read the peak. The readout shows the loudest frequency, its nearest musical note, and the current level — useful for spotting hum, tuning, or naming a pitch.
- 4Tune the analysis. Raise the FFT size for finer frequency resolution, or increase Smoothing to steady a jittery display. Lower both to react faster to quick sounds.
- 5Freeze to inspect. Press Freeze (or Space) to hold the current frame so you can read a peak or a spectrogram trail, then Resume.
How a real-time spectrum analyzer works
Sound reaches your ears as a single wobbling pressure wave over time — that's the time domain, and it's exactly what the oscilloscope view draws. A spectrum analyzer answers a different question: which frequencies make up that wave, and how loud is each? To find out it runs a Fast Fourier Transform (FFT) — a mathematical routine that decomposes a short slice of the waveform into a stack of pure sine components. The result is the frequency domain: the bar graph and the spectrogram are two ways of drawing it.
The FFT splits the range into evenly spaced bins. The number of bins is set by the FFT size, and each bin's width is the sample rate divided by the FFT size — so at 48 kHz, a 2048-point FFT gives bins about 23 Hz wide, while 8192 points narrows them to about 6 Hz. That's the core trade-off: a bigger FFT resolves closely spaced tones and low bass far better, but it needs a longer slice of audio, so the display reacts more slowly and smears fast transients. A smaller FFT is snappy and time-accurate but blurs neighboring frequencies together.
The bar view uses a logarithmic frequency axis because that's how we hear — each octave (a doubling of frequency) takes equal space, so the musically important low and mid ranges aren't crushed into the far left. Levels are shown in dBFS, decibels below digital full scale. The whole chain runs on the browser's Web Audio AnalyserNode from a live mic stream; the smoothing control simply blends each new frame with the last so the picture doesn't flicker.
What people use it for
Hunt electrical hum
Mains hum shows as a sharp, steady spike at 50 or 60 Hz (plus harmonics at 100/120, 150/180…). Spotting it pinpoints a ground-loop or interference problem.
Catch feedback and whistles
A rising, sustained narrow peak is the start of feedback or a resonant whistle. Find its frequency here, then notch or move the mic before it howls.
Judge room noise
Watch the noise floor with the room "silent" to see fridge hum, fan rumble, or HVAC energy — the broadband level the spectrogram settles at when nothing is playing.
Tune an instrument
Play a note and read the peak's nearest musical note and exact frequency — a quick sanity check against a proper tuner for guitar, voice, or synth.
Tame sibilance
Harsh "s" and "t" sounds cluster as bursts of energy around 5–9 kHz. Seeing where they land tells you where to de-ess a vocal.
Check frequency response
Feed a sweep or pink noise through speakers or headphones and watch which bands are strong or missing — a fast read on tone balance.
How to read what you see
The same shapes show up again and again. Here's what the common patterns mean:
A sharp spike at 50 or 60 Hz
Mains electrical hum picked up by a cable or a ground loop — 50 Hz in most of the world, 60 Hz in North America.
→ Look for evenly spaced harmonics above it. Reseat and reroute cables away from power supplies, and try a different outlet or a ground-loop isolator.
Evenly spaced spikes above a low peak
Harmonics of a musical note or tone — a fundamental plus overtones at 2×, 3×, 4× its frequency.
→ Nothing to fix — this is what pitched sound looks like. The lowest spike is the fundamental that sets the note you hear.
A single peak that keeps rising
Acoustic feedback or a resonance building at one frequency as a mic hears its own amplified output.
→ Note the frequency, then lower gain, move the mic away from the speaker, or notch that band with an equalizer.
A raised, flat noise floor
Broadband noise — hiss from high gain, a cheap preamp, or a noisy room — lifting the whole bottom of the display.
→ Lower input gain, move away from fans and HVAC, and enable noise suppression. A little floor is normal; a high one drowns quiet detail.
Bright energy around 5–9 kHz on speech
Sibilance — the hiss of "s", "sh", and "t" sounds concentrating in the high mids.
→ If it's harsh, angle the mic slightly off-axis and apply gentle de-essing in that band rather than cutting all the highs.
Lumpy peaks and dips in the low end
Room modes — standing waves where a room's dimensions reinforce or cancel certain bass frequencies.
→ Move the mic or speakers, or add bass trapping. This is a room acoustics issue, not a fault in the gear.
Spectrum analyzer glossary
- FFT
- Fast Fourier Transform — the algorithm that converts a slice of waveform (time domain) into its frequency content (frequency domain). It's what turns sound into the bar graph.
- Bin
- One frequency slot in the FFT output. Each bin covers a small band; its width equals the sample rate divided by the FFT size.
- Frequency resolution
- How finely the analyzer separates two close tones — the bin width. A larger FFT gives a smaller bin width and sharper resolution.
- dBFS
- Decibels relative to full scale. 0 dBFS is the maximum a digital signal can reach; everything quieter is a negative number, so levels here read below zero.
- Spectrogram
- A scrolling image of frequency over time, with color showing intensity. Reads left-to-right as history: bright trails are sustained tones.
- Oscilloscope
- A view of the raw waveform in the time domain — amplitude plotted against time — showing the shape of the wave rather than its frequencies.
- Fundamental
- The lowest frequency of a pitched sound, which sets the note you perceive. The peaks above it are its harmonics.
- Harmonic
- A frequency at a whole-number multiple of the fundamental. Their pattern and strength give an instrument or voice its timbre.
- Nyquist frequency
- Half the sample rate — the highest frequency that can be represented. At 48 kHz that's 24 kHz, the top of what the analyzer can show.
Frequently asked questions
What does a spectrum analyzer actually show?
It shows which frequencies are present in a sound and how loud each one is, right now. Where your ears and an oscilloscope see one combined waveform moving over time, the analyzer breaks that wave into its frequency ingredients using an FFT and plots them from low (left) to high (right). A low rumble sits on the left, a cymbal or an "s" sound on the right, and a pure tone appears as a single spike. It's the difference between hearing a chord and seeing each note that makes it up.
What's the difference between bars, oscilloscope, and spectrogram?
They're three views of the same live audio. Bars show the frequency spectrum at this instant — how much energy is in each band — and are best for spotting peaks, hum, and tone balance. The oscilloscope shows the raw waveform in the time domain, good for seeing the shape, period, and clipping of a wave. The spectrogram plots frequency vertically against time horizontally and scrolls, with brightness for loudness, so you can watch how a sound evolves — ideal for tracking a sweep, a melody, or a sustained whistle.
Which FFT size should I use?
It's a trade between frequency detail and speed. A large FFT (4096 or 8192) makes narrow bins, so it separates close tones and resolves bass far better — pick it for tuning, hum hunting, or examining a chord. A small FFT (1024) reacts faster and pinpoints quick transients like drum hits, but blurs neighboring frequencies together. 2048 is a sensible default. Because each bin's width is the sample rate divided by the FFT size, doubling the FFT halves the bin width and doubles the resolution.
How do I find electrical hum with this?
Play the suspect signal (or just let a humming device be heard) and switch to Bars or Spectrogram. Mains hum appears as a razor-sharp, rock-steady spike at 50 Hz in most countries or 60 Hz in North America, usually with smaller harmonics at 100/120 Hz, 150/180 Hz and up. A larger FFT size makes it easier to pin the exact frequency. Once identified, hum is almost always a cabling or grounding problem — reroute cables away from power bricks, try another outlet, or use a ground-loop isolator.
Can I use this to tune an instrument?
Yes, for a quick check. Play a single sustained note and read the peak frequency and its nearest note in the readout — the fundamental is the lowest strong spike. It's handy for confirming a string is roughly in tune or naming an unknown pitch. For serious tuning use a dedicated tuner or our tone generator as a reference pitch, since a spectrum analyzer reads to the bin width, not the fraction-of-a-cent accuracy a tuner is built for.
Is my audio recorded or sent anywhere?
No. The analyzer only reads the live microphone signal to compute and draw the spectrum — nothing is captured, saved, or uploaded, and there's no account or logging. The audio never even reaches your speakers, which is why there's no feedback. When you close or reload the page, the microphone is released and nothing remains. All processing happens on your device with the browser's built-in Web Audio API.
Why is the frequency axis not evenly spaced?
The bar view uses a logarithmic frequency axis, so each octave — a doubling of frequency — takes up the same width. We hear pitch that way: the jump from 100 to 200 Hz sounds like the same musical distance as 1,000 to 2,000 Hz, even though one spans 100 Hz and the other 1,000 Hz. A linear axis would cram all the musically busy low and mid content into the far left and waste the right half on barely audible highs, so a log axis matches perception and makes the display far more useful.