|Support type:||Scholarships in Brazil - Scientific Initiation|
|Effective date (Start):||July 01, 2016|
|Effective date (End):||April 30, 2018|
|Field of knowledge:||Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques|
|Principal Investigator:||Fabricio Aparecido Breve|
|Grantee:||Guilherme Rodrigues Vitorino|
|Home Institution:||Instituto de Geociências e Ciências Exatas (IGCE). Universidade Estadual Paulista (UNESP). Campus de Rio Claro. Rio Claro , SP, Brazil|
This work has an objective to realize audio processing with the proposal to reduce the possible noises in the signal, because these noises can be harmful to the understanding and utilization of the audio. Systems that uses audio signals can be victims of these noises that can prejudice the operation. Thus, many of them have a preprocessing step that performs an audio denoising before the utilization. Several algorithms were proposed along the years to reduce noise in audio signals. But the results oftentimes have generated files with high levels of distortion. In this way, this project consists in the development of new approaches to noise reduction in audio, by adapting methods in state-of-the-art of image denoising to audio noise filtering, Non-Local Means and Block-Matching and 3D filtering (BM3D), besides an approach based in Deep Learning. These approaches will be compared with traditional methods (Wiener Filter, for instance) and state-of-the-art techniques (of non-diagonal estimative) from the literature of filtering audio noise, besides methods in state of art of image denoising. Finally, it is expected that these new approaches can improve the signal-to-noise ratio of the audio signals provided to the system.