Desenvolvimento em HDL de uma alternativa de hardware dedicado para síntese de áudio utilizando o protocolo MIDI

Abstract

An audio synthesizer is a system able to play musical notes according to a determined pattern, mimicking real instruments. In musical theory, a note pitch is directly related to the fundamental harmonic frequency, as is an instrument characteristics (timbre) to the harmonics distribution, given that all harmonics are multiples in frequency of the first one. The Loris analyser models musical sounds decomposing them into harmonics, represented as frequencyand amplitude-variable in time sine waves. Based on this modelling, we propose a synthesizer, implemented in dedicated hardware, that recreates the sounds by modelling them from the multiple sine waves, based on communication using the MIDI protocol. The main challenge is the time aspect, because creating each one of the waves uses resource-consuming multiplication operations, and thousands of those should be generated in a few microseconds, because synthesizers are commonly evaluated according to the number of simultaneous notes they can play. The proposed system, limited by the programmable logic device in which it was tested, can generate up to 2048 simultaneous sine waves, enough for playing singleinstrument music correctly, with good performance and realistic enough sound quality, although some noise issues are perceptible. This result is a proof of the proposed concept and may base future work on adding new features and improving the already developed ones.