You should make different recognizers for calls that are significantly different, especially with limited training data. With a small data set and significant variation in call types, you are likely to create a recognizer that will only look for a very close match to each variation; the new recording is unlikely to fit within any one of the tight definitions and thus will fail to match.
You can run multiple recognizers at once if the recognizers are all in the same recognizer group – meaning – they are built with identical parameters, e.g. sample rate, FFT size, detector parameters, etc.
If you use data from only one individual, you are telling SongScope that a large number of individuals have very little variation in their calls. SongScope will reject calls that aren't nearly identical to this individual's calls. The solution is using more training data, preferably from different sites at different times and conditions to be sure that a dozen or so individuals are present. This will reflect the broadest range of variations present and avoid model segmentation.
As of November 2016, we have made Song Scope free. Simply download the latest version from our website, and you can use it without entering an activation code.
Please note that we have moved our focus to Kaleidoscope, which has capabilities including and beyond those of Song Scope, and consider Song Scope to be a legacy product with limited support. A free trial of Kaleidoscope is available here.