Spectrum Analyzer Basics

A spectrum analyzer is a tool that decomposes an audio signal into its constituent frequencies and displays their relative amplitudes. While a waveform shows amplitude over time (the time domain), a spectrum analyzer shows amplitude over frequency (the frequency domain). The most familiar visualization is a row of vertical bars, each representing a frequency band, bouncing in real time as music plays. Spectrum analyzers are used in audio engineering for mixing and mastering, in live sound for room equalization, and in music visualization as a core visual element. Understanding how they work provides insight into the fundamental relationship between audio signals and the visual representations used in every music visualizer.

The audible frequency range spans roughly 20 Hz to 20,000 Hz. This range is commonly divided into bands. Sub-bass (20-60 Hz) is felt more than heard and is produced by kick drums and bass synths. Bass (60-250 Hz) provides the low-end warmth of bass guitar, lower piano notes, and toms. Low-mids (250-500 Hz) contain the body of many instruments and can sound muddy if excessive. Mids (500-2000 Hz) are where most melodic content lives, including vocals and lead instruments. Upper-mids (2000-4000 Hz) carry vocal presence and the attack of guitars and snares. Presence (4000-6000 Hz) adds clarity and definition. Brilliance (6000-20000 Hz) includes sibilance, cymbals, and the airy quality of high-frequency content. Spectrum analyzers display the energy in each of these regions.

The Fast Fourier Transform (FFT) is the mathematical algorithm that converts time-domain audio samples into frequency-domain data. An FFT size of 2048 samples (at 44.1 kHz sample rate) produces 1024 frequency bins, each approximately 21.5 Hz wide. The lower bins resolve low frequencies, and the upper bins resolve high frequencies. However, human pitch perception is logarithmic: we perceive the interval between 100 Hz and 200 Hz as the same "distance" as between 1000 Hz and 2000 Hz (both are one octave). For this reason, spectrum analyzers typically group FFT bins on a logarithmic scale, allocating more visual space to lower frequencies and progressively grouping more bins together at higher frequencies. This produces a display that matches our perception of pitch.

The classic bar graph is just one way to display spectral data. A line spectrum connects the amplitude values with a smooth curve, producing a flowing mountain-like silhouette. A circular spectrum arranges frequency bands around a circle, with amplitude extending radially outward. A spectrogram (or waterfall display) shows frequency on one axis, time on the other, and amplitude as color, revealing how the spectrum evolves over the duration of the song. A 3D terrain visualization maps frequency to X position, time to Z position (depth), and amplitude to Y position (height), creating a landscape that you can orbit around. Each visualization type emphasizes different aspects of the audio: bar graphs are great for real-time monitoring, while spectrograms excel at showing long-term patterns.

In a music visualizer, spectrum data is one of the primary inputs driving visual behavior. A common technique is splitting the spectrum into a few broad bands (bass, mids, treble) and assigning each band to a different visual parameter. The bass band might control the scale of a central object, the mids might control color hue, and the treble might control particle emission rate. For more nuanced results, individual narrow bands can drive specific elements: the frequency range of a vocal formant might control a particular visual layer. Smoothing is essential to prevent jittery visuals: raw FFT data changes dramatically frame to frame, so exponential averaging or spring-based smoothing creates more pleasing motion. Zeltor Labs normalizes spectrum data per-song to ensure consistent visual behavior regardless of how a track is mastered.