![]() | Only 14 pages are availabe for public view |
Abstract Since the invention of the Fast Fourier Transform 1965, acousticians have found the digital methods for analysis increasingly attractive. This attractivness augmented by the commercial development of small digital processors suitable for laboratory, field and use. The need for a spectral representation which reflects the time varying properties of a waveform ( the speech for example) leads to the definition of a Short-Time Fourier Analysis. Several different procedures for spectral analysis can be described by the unified representation of the Short-Time Fourier Analysis. In this thesis, Two algorithms for digital spectral analysis are compared. The first is the conventional Discrete Fourier Transform (OFT). The second is the digital spectral analysis using the polyphase technique. The polyphase technique is an efficient structure of the Generalized Filter Bank based on the definition of the Short-Time Spectral Analysis. Close relations as well as differences between these two analyzers are demonstrated using the concept of the multirate systems. Simulation measurments’ of the chracteristics of both algorithms are carried on demonstrating the superpriority of the polyphase spectrum analyzer compared to the conventional OFT for more general signals. A hardware interface circuit is implemented to build a model of a real time digital spectrum analyzer based on a simulation of the polyphase structure. A practical digital analysis of deterministic signals and speech waveforms is carried on using this practical model. A comment on the theoretical and practical results is presented. (FFT) in spectrum has been dependable shipboard |