This spectrum analyzer splits the sound from your microphone into its frequencies and shows how loud each one is, live, on a logarithmic scale from 20 Hz to 20 kHz. Below it, a scrolling spectrogram shows the last few seconds as a colour map. It finds the strongest tones and names their musical notes, holds peaks, switches to one third octave bars like a hardware real-time analyzer, and checks for 50 or 60 Hz mains hum. It works with built-in laptop and phone mics, USB and headset mics and audio interfaces, in Chrome, Edge, Firefox and Safari on Windows, Mac, ChromeOS, Android and iPhone.
How to use the spectrum analyzer
- Press Start analyzer and allow the microphone. Pick another mic in Input if you have more than one.
- Make the sound you want to look at, or bring the mic close to it. Whistle or hum first to see how a single tone looks.
- Move the pointer, or your finger, across the spectrum to read the exact frequency, note and level under it.
- Press Freeze to stop the picture and study it, and Save image or CSV to keep it.
- Change FFT size, Averaging and Peak hold for the job: details for steady sounds, speed for changing ones.
What the graphs and readings show
| Reading | What it tells you | Good to know |
|---|---|---|
| Spectrum line (green) | The level of each frequency right now, in dBFS | 0 dBFS is the loudest the input can record, so readings are negative |
| Peak hold (yellow) | The highest level each frequency reached | Falling holds for 1 second, then drops 20 dB a second. Hold keeps it until you clear it |
| Spectrogram | How the spectrum changed over time, newest at the top | Brighter is louder |
| Dominant frequency and note | The strongest clear tone, with its note and how many cents it is off | Shown only when one tone stands well above the rest |
| Level | The level of the whole spectrum, with your weighting | Sound starts to distort as it nears 0 dBFS |
| Resolution | The gap between FFT bins, and the slice of time each picture covers | Sample rate divided by FFT size |
| Mains hum | Whether a 50 or 60 Hz tone with its harmonics is present | Needs FFT 8192 or bigger to tell 50 from 60 Hz |
FFT size, resolution and speed
At the usual 48 kHz sample rate, each FFT size trades detail against speed:
| FFT size | Resolution | Time slice | Best for |
|---|---|---|---|
| 2048 | 23.4 Hz | 43 ms | Speech, drums, fast changes |
| 4096 | 11.7 Hz | 85 ms | General music |
| 8192 | 5.9 Hz | 171 ms | Most uses, the default |
| 16k | 2.9 Hz | 341 ms | Hum, bass notes, close tones |
| 32k | 1.5 Hz | 683 ms | Very fine detail in steady sounds |
Averaging smooths the line over time, so noise looks steady and real peaks stand out. Weighting tilts the picture: A weighting turns down the bass and the very top the way your ears do at normal levels, C weighting only trims the extremes, and Z leaves it flat.
A low hum or buzz in your mic or speakers
A row of evenly spaced peaks at 50, 100, 150 Hz, or 60, 120, 180 Hz, is mains hum. The power grid runs at 50 Hz in most of Europe, Asia, Africa and Australia and at 60 Hz in North America and parts of South America and Asia. A buzz strongest at 100 or 120 Hz usually comes from a power supply. Fixes, in order:
- Unplug the laptop charger and listen again. If the hum stops, it is a ground loop through the charger. Plug the charger and the speakers or mixer into the same power strip.
- Move audio cables away from power cables, adapters and power strips.
- Use a ground loop isolator on a 3.5 mm or RCA cable between a computer and speakers, or balanced XLR or TRS cables for studio gear.
- Try another USB port, ideally one on the computer itself, for USB mics and audio interfaces.
A high whine from a computer or charger
A thin line high in the spectrum, often a few kilohertz, that rises and falls with what the computer is doing, is usually coil whine from a graphics card, power supply or charger. Watch the spectrogram while you start a game or a heavy task: if the line appears or moves with the load, the source is the computer. Capping the frame rate in games often moves or quiets it. A steady whine that stays put can be a fan bearing or a cheap USB charger.
Your mic sounds muffled or thin
Speak normally and look at where the spectrum falls away. A mic that drops off a cliff at about 8 kHz is running in a phone call mode: Bluetooth headsets switch to their hands-free calling profile when their mic is in use, which usually cuts everything above 8 kHz, or above 4 kHz on older headsets. Use a wired or USB mic for recording. A mic with almost nothing below 200 Hz sounds thin, which is normal for small laptop mics. Compare mics side by side with the mic quality test.
Feedback and ringing tones
When a mic picks up its own speaker, one frequency builds up and rings. Turn the volume up slowly with Peak hold on and watch for one narrow peak that grows faster than the rest. That frequency is the one to cut with a narrow equalizer band, or move the mic behind the speaker.
Boomy bass in a room
Rooms boost some bass frequencies and cancel others. Play pink noise through your speakers, switch to 1/3 octave with High averaging, and stand where you listen. Pink noise should come out roughly flat, so bands that stick up more than about 6 dB above their neighbours are room problems to treat with speaker placement or an equalizer. To hear a single frequency on its own, play it with the tone generator or step through the range with a frequency sweep.
Using it as a frequency analyzer for music and voice
Sing or play one note and the dominant frequency shows the note and how many cents sharp or flat it is. For singing practice with a moving pitch line, the pitch detector is easier to read. The strongest peaks list gives the first five tones in a sound, which helps find the key of a hum or match a tuning fork.
What a browser spectrum analyzer cannot show
The analyzer sees only what your microphone and sound card pass on. Many built-in mics pick up little below about 100 Hz and some fade out well before 20 kHz, so the very bottom and top of the graph often show the mic, not the sound. Levels are in dBFS, relative to the input, not in sound pressure: use the sound level meter for dB SPL readings. It cannot show anything above half the sample rate, which is 24 kHz at 48 kHz, and radio signals such as Wi-Fi and Bluetooth never reach a microphone at all.
Spectrum analyzer questions
What is an audio spectrum analyzer used for?
It shows how much of a sound sits at each frequency, from deep bass at 20 Hz to the highest treble at 20 kHz. People use it to find a hum or whine, check how a mic or speaker sounds, tune a room, set an equalizer, spot feedback before it squeals, and see the notes and harmonics in music or a voice.
What FFT size should I use?
Start with 8192. Go up to 16k or 32k to separate tones that are close together, such as 50 and 60 Hz hum or low bass notes. Go down to 2048 or 4096 for speech, drums and other sounds that change quickly, because a big FFT averages over a longer slice of time.
Can I analyze a song or a video instead of the mic?
The analyzer listens to an input device. On Windows, turn on Stereo Mix in Sound settings, then More sound settings, then the Recording tab, and pick it in Input to analyze what the computer is playing. Otherwise play the sound through a speaker next to the mic, which works for finding notes and rough balance.
Why do I see several peaks when I play one note?
Most sounds are made of a fundamental plus harmonics at 2, 3, 4 and more times its frequency. A note A3 at 220 Hz also shows peaks at 440, 660 and 880 Hz. The lowest of the evenly spaced peaks is the note you hear, and the pitch detector names it for you.
What does the spectrogram show?
Frequency runs across, matching the spectrum above it, and time runs down, with the newest sound at the top. Brighter colours mean louder: dark green is quiet, yellow and orange are loud and white is close to the loudest the input can take. A steady tone draws a vertical line, a whistle that slides up draws a slanted one.
Is a frequency analyzer the same as a spectrum analyzer?
For sound, yes. Both names describe a tool that splits a signal into its frequencies with a Fourier transform and shows the level of each. Radio spectrum analyzers work the same way at much higher frequencies, which a microphone cannot pick up, so Wi-Fi and Bluetooth signals will not show here.