Modeling Lemur Vocalizations from a Signal Processing Perspective

Most of the synthesis models for the generation of the animal vocalization until now have been studied with a non-probabilistic approach. Current studies are based on physical models, which have a large number of restrictions and limits for the creation of a realistic synthesis. In this thesis, we have created acoustic representations of black and white ruffed lemur (Varecia variegata) vocalizations, basing on signal processing techniques, and we have used them for achieving high-level synthesis. Afterwards, we have introduced Hidden Markov Models framework, approach which has been very successful in the context of speech synthesis. The outcome of the project has been evaluated by means of listening tests with humans, where original and synthetic vocalizations have been evaluated.