Spectrum

The fundamental question that the Spectrum answers is: at a given point in time, which frequencies are present in a sound, and how loud is each one? More precisely, for a given short period of time, what is the average intensity of each frequency over that period?

Basic Principles

Where the Waveform shows a sound in the time domain — amplitude against time — the Spectrum shows it in the frequency domain: intensity against frequency. The two views describe the same signal but answer different questions.

The transformation between the two views is performed by the Fast Fourier Transform (FFT). Given a short window of audio samples, the FFT computes how much energy is present at each frequency. The result is most revealing on stable, sustained sounds, where the spectrum settles into a clear pattern; on quickly changing material, each spectrum captures only a fleeting snapshot.

Reading the Spectrum

The horizontal axis shows frequency, typically up to about 5000 Hz for voice analysis; it can be set to a linear or logarithmic scale. The vertical axis shows intensity in decibels, with 0 dB at the top.

Peaks on the spectrum correspond to strong frequency components:

  • Sharp, narrow peaks at evenly spaced frequencies are harmonics — frequency components at integer multiples of a fundamental.

  • Broad peaks that envelop several harmonics are formants — resonances of the vocal tract that shape vowel sounds.

The relative heights of these peaks are what give a sound its timbre: the same fundamental sung as different vowels produces the same harmonic positions but different formant peaks.

Spectrum Types

VoceVista can compute the spectrum from windows of different lengths, and the choice has a strong effect on what the spectrum reveals.

Instantaneous Spectrum

Computed from a short window centred on a single point in time. Responds quickly to changes in the signal — the most detailed view, but also the most jittery.

Averaged Spectrum

Averaged over a longer window. Smooths out short-lived fluctuations and produces a more stable view of sustained sounds.

Long-Term Average Spectrum (LTAS)

Averaged over a longer section of the recording — often the whole file — to reveal the overall spectral fingerprint of a voice or instrument. See Long-Term Average Spectrum for more.

Applications in Voice Science

For singing teachers and voice scientists, the Spectrum makes objective measurements available for phenomena that are otherwise described only subjectively. From a single sustained tone you can read:

  • the fundamental frequency (the pitch of the note);

  • the distribution of energy among the harmonics, which determines the brightness or warmth of the timbre;

  • the formants, in particular the strength and position of the singer’s formant in trained classical voices;

  • the effect of different vowels and different vocal techniques on the resonance pattern.

In VoceVista

The Spectrum is part of the Analyzer View. FFT size, window function, and the choice between instantaneous and averaged spectra are configured in Analyzer Settings; for the dedicated long-term display, see Long-Term Average Spectrum.