Audio Spectrum Analyzer — Free

This online audio spectrum analyzer transforms incoming sound into a live, logarithmic frequency-domain display spanning 20 Hz to 20,000 Hz. Built for audio engineers, acoustic technicians, and producers, it lets you identify resonant room peaks, detect electrical ground hum, evaluate stereo frequency balance, and analyze harmonic overtone structures in real time using client-side Fast Fourier Transform processing.

How to use it

1. Select your audio input source. Start with the built-in 1000 Hz reference sine generator to verify analyzer calibration, or select an external microphone or audio interface from the device menu.
2. Observe logarithmic frequency bars. The horizontal axis arranges frequencies logarithmically from 20 Hz to 20,000 Hz so each musical octave occupies equal screen width. Vertical bar height indicates instantaneous signal level in dBFS.
3. Isolate stereo channels. Toggle between the Left channel (deep sky blue), Right channel (warm amber), or dual L+R mode to compare spectral balance across both sides of a stereo mix.
4. Enable Peak Hold. Activate peak hold markers to retain a thin horizontal indicator at the maximum recorded level for each frequency band, highlighting brief transient spikes.
5. Engage the scrolling spectrogram. Toggle the waterfall spectrogram to display frequency energy as a continuous heat map scrolling downward over time, making resonance and frequency drift visible.
6. Freeze the display with Pause. Click Pause to halt data updates without stopping audio playback. Freezing lets you read exact frequency values in hertz and amplitude levels in dBFS via the text readout.

Reading the display and spectral acoustics

A spectrum analyzer computes the Fast Fourier Transform (FFT) of discrete audio buffers, converting time-domain sound into discrete frequency components. While an oscilloscope displays amplitude over time, a spectrum analyzer displays energy across the acoustic spectrum.

Human hearing perceives pitch logarithmically. Musical octaves represent successive doublings of frequency: 100 Hz to 200 Hz spans one octave, while 10,000 Hz to 20,000 Hz also spans one octave. A linear scale would compress 90% of musical information into the first tenth of the screen; a logarithmic scale allocates equal visual space to each octave.

Real-time analyzers group raw FFT bins into fractional octave bands, typically 1/3-octave intervals conforming to ISO 266:1997 preferred frequencies. This fractional grouping mirrors the critical bands of the human ear, where filter bandwidth broadens at higher frequencies.

Distinct acoustic phenomena produce characteristic signatures on the display:

• Electrical mains hum manifests as a sharp spike at 50 Hz or 60 Hz. Imperfect grounding also generates odd-order harmonic spikes at 150 Hz and 180 Hz.
• Room resonance modes appear as broad peaks between 30 Hz and 200 Hz. Unlike narrow mains hum, room modes excite clusters of bins whose levels change as you move through the space.
• Microphone proximity effect causes a 6 dB to 12 dB boost below 200 Hz when directional cardioid capsules are placed within 5 cm of a source.
• Vocal sibilance concentrates energy between 5000 Hz and 8000 Hz, showing where harsh consonant friction (“s”, “sh”, “t”) requires de-essing or microphone adjustment.

Technical specifications and reference values

The table below details standardized 1/3-octave center frequencies per ISO 266:1997 and filter bandwidths conforming to IEC 61260-1 standards, along with characteristic content and diagnostic applications.

Acoustic Band	Center Frequency (Hz)	Bandpass Limits (Hz)	Characteristic Audio Content	Acoustic Diagnostic Focus
Deep Sub-Bass	31.5 Hz	28.1 – 35.5 Hz	Pipe organ low pedal, sub-kick rumble	Subwoofer extension, structural room modes
Sub-Bass	63 Hz	56.2 – 70.8 Hz	Kick drum body, 60 Hz electrical mains hum	Mains hum detection, bass reflex port resonance
Mid-Bass	125 Hz	112 – 141 Hz	Bass guitar punch, male vocal fundamental	Chest resonance, lower boundary standing waves
Lower Midrange	250 Hz	224 – 282 Hz	Guitar body warmth, snare drum fundamental	Boxy acoustic build-up in untreated rooms
Midrange	500 Hz	447 – 562 Hz	Vocal body, orchestral brass warmth	Speech intelligibility warmth, comb filtering
Reference Midrange	1000 Hz	891 – 1122 Hz	Standard audio test tone (ISO 266 reference)	System gain staging benchmark, converter linearity
Upper Midrange	2000 Hz	1782 – 2245 Hz	Speech articulation, electric guitar attack	Perceived loudness, converter harshness
Presence	4000 Hz	3564 – 4490 Hz	Vocal consonant snap, snare transient crack	Ear canal acoustic resonance peak (ISO 226)
Brilliance	8000 Hz	7127 – 8980 Hz	Cymbal sizzle, vocal breath, acoustic sparkle	High-frequency driver roll-off, preamp tape hiss
Ultra-High Brilliance	16000 Hz	14254 – 17959 Hz	Air sheen, extreme upper overtones	Anti-aliasing filter cutoff, adult hearing limits

Accuracy, limits and troubleshooting

Browser-based spectrum analysis relies on the Web Audio API AnalyserNode. The browser slices streaming audio into discrete time windows and computes a Fast Fourier Transform. Understanding the technical boundaries of FFT mathematics prevents misinterpreting measurement data.

The time-frequency resolution trade-off is governed by the Heisenberg-Gabor uncertainty principle. The frequency resolution of an FFT is determined by $\Delta f = f_s / N$, where $f_s$ is the audio sampling rate and $N$ is the FFT buffer size. At a standard 48,000 Hz sample rate with an FFT size of 2048 samples, the resulting bin width is approximately 23.44 Hz ($48000 / 2048$). While 23.44 Hz provides fine resolution at 5000 Hz, it provides coarse resolution at 40 Hz, where consecutive bins jump from 23.4 Hz to 46.9 Hz. Increasing buffer size to 8192 yields a finer 5.86 Hz bin spacing but introduces a 170.67 ms buffer delay that slows transient visual response.

Window functions prevent spectral leakage. Slicing audio into finite buffer frames creates artificial discontinuities at buffer edges unless the wave crosses zero precisely at both ends. This step discontinuity introduces false high-frequency energy across the entire spectrum. The analyzer applies a windowing function, such as a Hann or Blackman-Harris window, tapering buffer edges to zero. This suppresses sidelobe leakage but slightly broadens individual spectral peaks.

Microphone hardware directly colors incoming acoustic readings. Standard internal laptop and mobile microphones are uncalibrated MEMS capsules with high-pass filters rolling off sharply below 80 Hz and resonance peaks between 6 kHz and 12 kHz. Using an uncalibrated microphone shows artificially suppressed bass and exaggerated treble compared to true room sound pressure levels.

Operating system audio processing alters input spectra. Features like echo cancellation and noise suppression attenuate steady-state background tones. When testing room noise or continuous test sweeps, disable automatic gain control and noise suppression in operating system sound settings to evaluate the true physical signal.

Related tools

• Audio Oscilloscope — View raw time-domain waveform curves, zero-crossings, and peak digital clipping in real time.
• Tone Generator — Produce calibrated sine waves from 20 Hz to 20,000 Hz to verify individual analyzer frequency bins.
• Decibel Meter — Measure sound pressure levels with standardized A-weighting and C-weighting acoustic filters.
• Room Acoustic Test — Identify standing room modes, acoustic nulls, and flutter echoes across your listening space.