Izračun spektra diskretnega signala: primerjava izvedb s CPU in GPU na primeru premika višine tona

Milharčič, Timotej (author), Burnik, Urban (mentor)

Delo obravnava časovno in frekvenčno analizo zvočnih signalov s poudarkom na hitri Fourierevi transformaciji (FFT), praktični implementaciji, aplikacijah in analizi učinkovitosti. Delo se začne s teoretičnim uvodom v Fouriereve algoritme, kjer so predstavljene različne oblike transformacije (integralna, diskretna, hitra) in njih inverzne transformacije. Osrednji del je namenjen razlagi in uporabi FFT algoritma, ki se v nadaljevanju osredotoča na dve praktični aplikaciji: - premik višine tona - sprotna (realno časovna) obdelava signala Zaključni del je namenjen primerjalni analizi hitrosti različnih FFT algoritmov. Primerjamo zmogljivosti algoritmov, ki se izvajajo na centralni procesni enoti (CPE), grafični procesni enoti (GPE) in v kombiniranih načinih, s čimer ocenimo njihovo učinkovitost za sprotno obdelavo signala.

• Language: Slovenian
• Keywords: Premik Tona, Sprotna obdelava, FFT, CPE, GPE, Primerjalna analiza
• Work type: Bachelor thesis/paper
• Typology: 2.11 - Undergraduate Thesis
• Organization: FE - Faculty of Electrical Engineering
• Year: 2025
• PID: 20.500.12556/RUL-171951
• COBISS.SI-ID: 263476483
• Publication date in RUL: 04.09.2025

Secondary language

• Language: English
• Title: Spectrum Calculation of a Discrete Signal: CPU and GPU Implementation Comparison with Pitch Shifting

Abstract

The paper deals with the time and frequency analysis of audio signals with an emphasis on a fast Fourier transform (FFT), practical implementations, applications, and performance analysis. The paper begins with a theoretical introduction to Fourier algorithms, presenting various forms of transformations (integral, discrete, fast) and their inverse transformations. The central part is devoted to the explanation and application of the FFT algorithm, which then focuses on two practical applications: - pitch shift - real-time signal processing We devote the final part to a comparative analysis of the speed of different FFT algorithms. We compare the performance of algorithms running on the central processing unit (CPU), graphics processing unit (GPU), and in combined modes to assess their effectiveness for real-time signal processing.

• Keywords: Pitch shift, Real-time processing, FFT, CPU, GPU, Comparative analysis