Software based analysis and synthesis of audio signals

U radu se objašnjavaju pojmovi vezani za zvučne signale, posebice glasovne te njihove značajke poput amplitude, spektralne ovojnice i formanta. Predstavlja se anatomski model vokalnog trakta i objašnjava kako se odvija cijeli proces proizvodnje glasa, od pluća koja u suradnji s glasnicama služe kao generator zvučnog signala, do vokalnog trakta koji modulira zvučne valove i daje im specifične značajke. Potom, preko nekoliko iteracija mehaničkih modela za proizvodnju glasa, dolazimo do računalnih modela koji sintetiziraju glas te prepreka koje postoje u tom pristupu. Konačno, prikazana je usporedna analiza sintetiziranog i prirodnog glasa.This paper explains the basics of audio signals, especially voice signals and their features such as amplitude, spectral envelope and formants. An anatomical model of the vocal tract is present, and the complete voice production process is explained, from the lungs which alongside the vocal cords serve as the sound wave generator, to the vocal tract which modulates the sound waves and gives them their specific features. Then, after a few iterations of mechanical models for voice production, we arrive at the modern voice synthesis computer models and the obstacles that follow. Finally, a comparative analysis of the synthesised and natural voice is shown