Digitalna obrada signala u glazbi

U ovom radu dan je pregled osnovnih pojmova i metoda vezanih uz digitalnu obradu signala. Definiran je pojam signala i navedene su razne podjele. Reprezentacija signala u digitalnoj domeni moguća je u velikoj mjeri zbog ograničenja ljudskog sluha. Stoga je opisana čovjekova percepcija psihoakustičnih svojstava signala: visine, glasnoće i boje te njihova veza s fizičkim svojstvima signala, frekvencijom i intenzitetom. U digitalnim sustavima možemo raditi jedino s konačnim brojem podataka zbog čega je potrebno vremenski kontinuirani analogni signal modificirati da u konačno mnogo podataka što vjernije reprezentira original. Opisani su uzorkovanje, kvantizacija i kodiranje - glavni postupci pri pretvaranju analognog u digitalni signal, te nusprodukti do kojih može doći zbog korištenja tih postupaka, poput preklapanja spektra. Dan je opis pulsno-kodne modulacije, jedne od arhitektura sklopova za pretvaranje analognog u digitalni signal. Navedena je osnovna terminologija i koncepti vezani uz filtere u frekvencijskoj i vremenskoj domeni. Promatran je osrednjavajući filter uz primjer implementacije u programskom jeziku Python.In this thesis an overview is given of basic terms and methods regarding digital signal processing. The term signal is defined and various categories of signals are mentioned. Representation of signals in the digital domain is in large part possible due to deficiencies in human hearing system. Hence, human perception of three major psychoacustic signal properties: pitch, loudness and color, is described, along with their connection to the physical signal properties frequency and intensity. In digital systems one can only work with a finite amount of data and storage. That is why it is necessary to modify time continuous analog signals to a finite amount of data which represents the original signal as closely as possible. Main methods in converting time continuous analog signals to discrete time digital signals: sampling, quantization and coding, are described, along with some of the byproducts, like aliasing. The description of a pulse-code modulation architecture is given. Basic terminology and concepts regarding filters in frequency and time domain are stated. Moving average filter is described and Python implementation of a said filter is given