Timbre & spectra / A listening lesson

A spectrum shows what is inside a sound

Read frequency and level without mistaking a chart for the entire listening experience.

A different view of the same signal

A waveform presents a signal over time. A spectrum presents its frequency content over a specified analysis interval. A horizontal position usually means frequency in hertz; height may mean amplitude, power or a logarithmic level. Read those labels before comparing graphs. A bar twice as tall on a linear amplitude axis does not mean the same thing as a six-decibel increase on a dB axis.

Evidence: 1

Distinguish calculated ingredients from measurement

A harmonic builder can list the exact frequencies and coefficients that it sends to synthesis. That is an analytical component spectrum. A live analyser instead estimates frequency content from a finite block of samples. Peaks may spread across neighbouring bins, and low components can be hidden by the chosen range. Neither display measures your loudspeaker or listening room without a microphone, which this site does not use.

Evidence: 2

Window length changes the question

A longer analysis window can separate nearby steady frequencies more finely, but combines information from a longer stretch of time. A short window follows changes quickly while giving a coarser frequency description. Attack sounds and beating therefore benefit from reading waveform, time values and spectrum together rather than asking one still image to explain everything.

Evidence: 3

Try a prediction before listening

Set the fundamental to 250 Hz and enable harmonics one, two and four. Predict spectral components at 250, 500 and 1,000 Hz. Then inspect the display and listen. Change only the fundamental to 300 Hz; the corresponding components become 300, 600 and 1,200 Hz. Change only a coefficient instead and the frequency positions remain fixed while their relative strengths change.

Evidence: 4

Analysis resolution

For N samples at sample rate fₛ, discrete Fourier transform bin spacing is fₛ/N. Windowing and signal duration influence actual peak resolution; bin spacing alone is not a measurement-accuracy guarantee.

Keep this in mind

A spectrum needs labelled units, an analysis interval and a distinction between a model and a measurement.

Make the connection

Build a sound

Predict the component frequencies before enabling harmonics 1, 2 and 4.

Open the experiment ↗

Useful words

Spectrum

A representation of a signal’s components across frequency. Its meaning depends on the time window, analysis and magnitude or power scale used.

See in glossary ↗

Sources & further reading

These sources support the explanations and model assumptions. Recording credits are listed in the listening panels and the source register.

  1. What is a sound spectrum? ↗Joe Wolfe / UNSW Music Acoustics · Undated
  2. Time and frequency ↗Chris Donahue / Carnegie Mellon University · 2026
  3. Amplitude and phase spectra ↗Chris Donahue / Carnegie Mellon University · 2026
  4. Additive synthesis ↗Chris Donahue / Carnegie Mellon University · 2026
Full source and rights register ↗