What is Overlap?
Overlap is a way of improving the time resolution of the spectrogram. From what described in Frequency resolution and time resolution, the time resolution dt is equal to N/fs, that is, the longer the SFT frame, the longer dt, the worse the time resolution. This means that events occurring within time t to time t+dt fall into one “column” in the spectrogram and cannot be distinguished from each other. Overlapping helps with this, like rewinding a bit and listening again.
Below: Difference between non-overlap and overlap processing of a signal transformed into Fourier transform frames. a) No overlap. b) 75% Overlap. Note that in b) four times as many Fourier transforms are performed, increasing the time resolution by a factor of 4. SFT means Short-Time Fourier Transform.
When Overlap is used, an event that does not even last as long as one SFT frame can be “seen” (even if at reduced amplitude) in many sets of spectra that are displayed adjacent to each other. This way visibility of very short temporal variations that occur in a signal is augmented.
The disadvantage of Overlap is that a higher value results in longer processing time when you let UltraVox detect calls.
Effect of Overlap on your spectrogram
Overlap has an effect on the spectrogram. The width of the spectrogram pixel is:
nWhen Overlap (%)= 0, it is the time resolution dt according to the table which you can find in Frequency resolution and time resolution.
nWhen Overlap (%)= 50, it is 50% of the value in the table.
nWhen Overlap (%)= 90, it is 10% of the value in the table.
Below: Spectrogram resolution when Overlap (%)= 0 (left), 50 (middle), and 90 (right). In this example, SFT length = 256, therefore dt=1.024 ms.