Dear all, We are proceeding with the algorithm which is going to filter the data to be annotated. It works quite well for goats and young goats. The problem arises with the sneezing sounds which are very similar with background noises. Please see some examples to see how alike these are in the following link https://we.tl/t-CG92O2COLl. It is really difficult even for humans to understand the difference between sneezing and background noise. So we need to think how important the sneezing class is, at least for this stage of the project where we are trying to annotate the available data. If you agree, we can discard sneezing for now and continue with the annotation of the rest interesting classes. And then at a later stage, we can try to improve the sneezing detection if you think it is important. Let us know you opinion on this. We may also set a meeting to talk about this. best regards, Stavros Ntalampiras -- *********************************** Dr.-Eng. Stavros Ntalampiras Associate Professor, Department of Computer science University of Milan, Italy https://sites.google.com/site/stavrosntalampiras/home ***********************************