now get some footage of their communications, and let's finally put that to some unsupervised learning algorithm so it distills some patterns in their audio/vi

now get some footage of their communications, and let's finally put that to some unsupervised learning algorithm so it distills some patterns in their audio/visual communication and then builds parrot2vec. Then you perform clusterisation analysis, and obtain some characteristic patterns. At least we'll have the vocabulary size with some precision. The vocabulary of bored domestic animal, therefore reduced to some degree..