Audio Spectrum Analyzer

Q: Is an audio spectrum analyzer different from a frequency analyzer?

No — same measurement, different emphasis. Spectrum analyzer leans engineering, frequency analyzer leans general; one live tool, two entry pages.

Analyze spectrum live ↓ Frequency Analyzer

The live analyzer below runs on the same on-device engine as the Frequency Analyzer — engineering preset: spectrum view with Z-weighting (flat) for raw diagnosis.

What an Audio Spectrum Analyzer Shows

Instead of collapsing everything into "70 dB", the spectrum analyzer plots dB per frequency band as vertical bars. Peaks reveal the dominant center frequencies: a 60 Hz hum, a 1 kHz tone, speech energy around 2–3 kHz, or broadband traffic hiss. It is the same FFT path described in Methodology — Spectrum (2048-point FFT, 64 grouped bars) with the same A/C/Z weighting correction that shapes what the bars emphasize.

Spectrum Analyzer vs Frequency Analyzer — Which Should You Use?

Both pages point to the same live FFT visualization, but they serve different readers. Choose this page if you want engineering detail — resonances, rumble-vs-hiss diagnosis, weighting comparison, and FFT limits. Choose the Frequency Analyzer page if you want a gentler introduction to bass, mids, and treble. The canonical live meter is at /#meter; each page is a distinct guide around it, not a duplicate.

Differentiation: this page frames the spectrum for engineering and source diagnosis (resonances, rumble vs. hiss, weighting comparison, export workflow); the Frequency Analyzer page frames it for general spectrum literacy (bass/mids/treble intuition with an illustrated example). Same on-device engine, complementary explanations — each guide stands alone for its audience.

How to Use the Analyzer

Start measuring on the meter at a stable distance from the source.
Open Visualization → Spectrum. Watch which third of the display lights up most — left (bass), center (mids), right (treble).
Toggle Z for raw physics, A for perceived-loudness weighting, C for loud bass-heavy setups. Compare with dBC vs dBA.
Freeze via Pause to photograph or describe a peak; Stop preserves statistics for export (CSV/JSON/PDF).

Engineering Notes

FFT size: 2048 → ~1024 bins to Nyquist. Grouped into 64 displayed bars; per-bin Δf ≈ sampleRate/2048 (~23 Hz @ 48 kHz). Fine for pattern recognition, not for resolving adjacent tones.
Windowing & smoothing: the underlying analyser applies smoothing per time-weighting (Fast ~0.2, Slow ~0.65, Impulse ~0.08). For repeatable band comparison, use a steady tone and Slow.
Mic bias: consumer mics roll off lows/highs; treat absolute bar heights as estimates. Relative comparisons (e.g. "peak moves from 120 Hz to 2 kHz when the speaker changes") are robust.
Export: CSV/JSON/PDF exports preserve the session statistics and history; the spectrum image can be captured via screenshot or included in the PDF snapshot.

FAQ — Audio Spectrum Analyzer

Is an audio spectrum analyzer different from a frequency analyzer?

No — same measurement, different emphasis. "Audio spectrum analyzer" leans engineering (visual spectrum with bars); "frequency analyzer" leans general (bass/mids/treble breakdown). This site treats them as one live tool with two entry pages.

Can I see Hz values on the spectrum?

The live display is grouped into bands for readability rather than per-Hz labels. Left is bass (~20–250 Hz), center is mids/speech (~250–4000 Hz), right is treble. For note identification, compare bar positions and switch weighting.

How accurate is the spectrum on a phone?

Useful for pattern and comparison, not lab-grade band levels. Absolute levels at extremes of the audible range are most affected by the phone's frequency response and any remaining OS processing.

Related: Frequency Analyzer · dBA · dBC · Leq Meter · Methodology