This thesis examines possibility of building guitar amplifier model with machine learning models and black box approach. It presents automated signal recording with its conversion and pre-processing for machine learning models. Signals where processed in frequency and time domain. Different algorithms emphasised on neural network and random forest were studied. It describes different points of view and approaches to the addressed problem. Problem was studied throughout working zone with sine signals, on single setting with sine signals and on single setting with guitar signals. Although considered on different working zones and domains, the hypothesis, that it is possible to build quality guitar amplifier with machine learning and black box approach, were refuted.
|
