Audio Spectrum Analyzer

Visualize an audio file or your live microphone input as a waveform, a logarithmic frequency spectrum, or a scrolling spectrogram. Everything runs locally in your browser — nothing is uploaded. Export any view as a PNG snapshot.
No audio loaded.
Ready.
What is an audio spectrum?

A waveform tells you how loud audio is at each moment; a spectrum tells you which frequencies make up that loudness. The analyzer above offers both views plus a spectrogram, which is the spectrum evolving over time. Same signal, three complementary pictures — and switching between them is the fastest way to understand what a sound is made of.

Four facts to anchor the rest of the page:
• waveform = loudness over time; spectrum = frequency content at one instant; spectrogram = spectrum over time (time × frequency × intensity)
• the math behind the spectrum is the FFT, applied to short windows of samples
• a 44,100 Hz file can only contain frequencies up to 22,050 Hz (half the sample rate — the Nyquist limit)
• the log frequency axis option matches how hearing works: the octave from 20–40 Hz gets the same width as 10,000–20,000 Hz
The three views

The waveform plots sample amplitude against time — excellent for spotting clipping, silences and overall dynamics, blind to pitch. The spectrum takes one short window (a few thousand samples) and shows how much energy sits at each frequency band, from the lowest bass on the left to the treble ceiling on the right. The spectrogram stacks hundreds of those spectra left to right as time passes, with intensity as colour — a piano glissando becomes a rising streak, a sustained violin note a steady horizontal line, and speech a chain of vowel formants.

FFT and windows

The Fast Fourier Transform converts a block of time-domain samples into frequency magnitudes. It answers: "if this block were a sum of pure tones, how much of each tone would it take?" The analyzer cuts the audio into overlapping windows of a few thousand samples and runs an FFT on each — smaller windows resolve timing better, larger windows resolve frequency better, which is why a fixed window size is always a compromise. The log-frequency toggle above is purely display-side: it stretches the plot so the low octaves (where most music lives) get visible width instead of being crushed into the left edge.

Fundamental and harmonics

Play an A4 on any instrument and the spectrum shows a peak at 440 Hz — the fundamental — plus smaller peaks at 880, 1320, 1760 Hz: the harmonics, whole-number multiples of the fundamental. It is the balance of those harmonics that makes a violin's A4 sound different from a flute's: same fundamental, different recipe above it. This is why the spectrum is the go-to view for instrument identification, mixing decisions and noise hunting — a mains hum shows as a stubborn 50/60 Hz spike, and MP3 truncation shows as energy that simply stops below about 16 kHz.

Common misconceptions
  • The waveform shows pitch. No. Two very different chords at the same volume have nearly identical waveforms. Pitch information only appears after the FFT — in the spectrum or spectrogram views.
  • A spectrum and a spectrogram are the same picture. No. The spectrum is a single instant's frequency slice; the spectrogram is thousands of those slices sequenced in time. Use the spectrum to answer "what", the spectrogram to answer "when".
  • Audio above 22,050 Hz is hidden in a CD file. No. Sampling at 44,100 Hz mathematically caps the content at 22,050 Hz; the Nyquist limit is a property of the samples themselves, not of the display.

Related tools: MIDI to WAV and Audio to WAV / FLAC to produce files worth analysing, Noise Reduce when the spectrogram reveals steady hiss or hum, and Audio Normalize when the waveform shows levels that need correction.