The Development, Implementation and Analysis of a Real-Time Parallel Algorithm of Sliding Discrete Fourier Transform

  • Iryna Tsimashenka

Student thesis: Masters ThesisMPhil

Abstract

Audio processing is an interesting and challenging area due to the strict requirements of highlevel human ear perception. Although audio processing is a part of digital signal processing, there has not yet been any realtime parallel implementation of the essential signal processing tool. We have developed, implemented and analysed a Sliding Discrete Fourier Transform algorithm using a particular vector parallel processing. This investigation is focused on speeding up realtime parallel implementation of the SDFT algorithm. It has developed and implemented some realtime parallel algorithms with different methods of data copy in order to find the fastest algorithm. In order to achieve a whole signal processing tool for professional use in the audio processing field a realtime parallel algorithm of Inverse Discrete Fourier Transform has been developed. In order to speed up this algorithm some versions were developed with different methods of data copy. All algorithms were tested and analysed. Realtime parallel algorithms were achieved with promising results.  The signal processing tool of the SDFT and the IDFT algorithms was attained.
Date of Award1 Jan 2011
Original languageEnglish
Awarding Institution
  • University of Bath
SupervisorJohn Fitch (Supervisor) & Russell Bradford (Supervisor)

Keywords

  • sliding discrete Fourier transform parallel algorithm
  • clearspeed accelerator

Cite this

'