access_time 21 de janeiro de 2020 às 11:00 até 21 de janeiro de 2020 às 12:00
place Room 2-N7.1, IST, Taguspark
Speech and face recognition are some of the existing tasks that are solved more quickly by humans rather than computers. The past years have witnessed substantial progress toward Artificial Intelligence but more concretely in regards to the speech. The traditional automatic speech recognition is a true example that represent words as sequence of phonemes by breaking the problems into keys components. However, for those that are new in speech recognition field and the related areas sometimes can be tough to deal with some concepts. The current work, besides demonstrating how speech can be processed since the speech signal processing, features extraction techniques to the interpretation of the results, in general presents an overview of existing literatures, most concretely the widely used models and their core elements such as the perceptron, neural networks, recurrent neural networks. We conclude this work with an short discussion of future directions.
local_offer Tópicos de Investigação
person Candidato: Frederico Samuel Felisberto
supervisor_account Orientador: Prof. Andreas Miroslaus Wichert