Analyzer Settings

The frequency resolution is the smallest difference (in Hz) between two frequencies that the analyzer can distinguish. Internally this is controlled by the FFT size — the number of points fed to each transform — but the dialog presents it as a frequency value in Hertz.

For a broader overview of what high FFT resolution unlocks and how it relates to sampling rate and microphone bandwidth, see the high-resolution audio spectrum analyzer page.

The time window determines how much audio is folded into each FFT update — and therefore the trade-off between time resolution and frequency accuracy. A longer window improves frequency accuracy; a shorter window shows faster temporal changes.

The dropdown selects how you specify the window:

For example, with a 44100 Hz sampling rate and an FFT size of 8192 points, the maximum window is approximately 185.8 ms — about 5.4 updates per second.

The short-term spectrum is averaged over this sliding time window during recording and playback to reduce flickering. Shorter durations make the spectrum more responsive to sudden changes but flicker more; longer durations are more stable but slower to react.

The Long-Term Average Spectrum (LTAS) is computed by averaging many spectral slices taken across a selected time range. This control sets how many slices per second are taken. Higher values yield smoother LTAS results at the cost of more computation.

The FFT method requires its input to be filtered through a window function to avoid spectral leakage. A discussion can be found at en.wikipedia.org/wiki/Window_function.

For VoceVista the default Dolph-Chebyshev window works very well and gives a usable dynamic range of about 100 decibels. Do not change it unless you understand window functions and their respective trade-offs.

The length of the audio segment fed to the pitch algorithm. The combo box shows both the number of samples and the resulting minimum detectable frequency (a 1024-sample window at 44100 Hz reaches down to about 43.1 Hz).

Longer time windows extend the algorithm’s reach to lower frequencies but reduce how quickly it can track changes. The default is appropriate for most situations.

The minimum confidence the algorithm must reach to accept a pitch result as valid, expressed as a percentage. Higher values reject ambiguous detections; lower values accept more — at the cost of more false positives.

The minimum signal level (in dB) required for a pitch reading to count. Signals quieter than this threshold are ignored, even when their clarity passes the test above. This keeps background noise from producing spurious pitch tracks during silent passages.

When checked, the pitch detector tries to find the fundamental frequency of harmonic sounds rather than reporting whichever harmonic is loudest. This is the right choice for analyzing musical instruments and the singing voice, where the fundamental is often weaker than its upper harmonics.

VoceVista ships with four built-in profiles, tuned for different kinds of material:

The Manage Profile button next to the profile dropdown opens a menu with the available actions:

Switching profiles applies all of that profile’s settings at once, so profiles are the fastest way to retune the analyzer for a different kind of material.