I'm calculating soundscape indices using scikit-maad's all_temporal_alpha_indices
and all_spectral_alpha_indices
. I will be using them as a "screen" to help identify areas of interest in hundreds or thousands of hours of acoustic recording. So was thinking to partition the recording into 1-minute windows and calculate the indices to 1 minute at a time.
Is there a minimum duration that the functions will work well for? Is 1 minute too short?
Or even if 1 minute is fine, the last window of a file might be incomplete, i.e. not a full minute. In that case I was thinking to still calculate the indices on the shorter window, unless there's some min duration under which the indices are not meaningful.