rafael-alvarez
commented
9 years ago a recording that only contains the species sound, does not contribute valuable information to the model
the vector computed from a recording with only the species sound, will be a vector of length 1
some statistics can not be computed from a vector of length 1 (standard deviation cannot be computed for example)
user having problem using validations with short recordings having only the sound. After discussing looking the code with @g-i-o-, we think this should be allowed and limit the recording size to the size of the pattern instead of 2 times the size of the pattern that is the limit now.
how this will affect the model and what you think @rafael-alvarez ??