Abstract
This work investigates brass instruments in the time domain, rather than the traditional frequency domain, and considers first, impulse measurements and secondly, their analysis. An existing apparatus for measuring the response to an acoustic impulse at the input of a brass instrument has been refined. Problems of impulse inconsistency, ambient temperature variation and source reflections have been resolved. Developments of the above equipment are used to test the quality of brass instruments on a factory production line. A prototype and a test instrument are compared by taking the arithmetical difference of their impulse responses. The equipment has detected small faults missed by normal inspection methods. The usefulness of this technique to brass instrument manufacturers is discussed. Links between the instrument's measured transient response and its bore geometry have been developed. The stages involved are deconvolution and bore reconstruction. Various deconvolution methods have been studied systematicaly by applying them to simulated noiseless and noisy data. Noise introduces errors, particularly at high frequencies, so deconvolution of real measured data is distorted. Techniques to reduce the effects of noise have been investigated. Attempts to employ the Gerchberg restoration algorithm 'to restore high frequency information proved unsuccessful. A new inverse method, based on an iterative z-transform procedure, of reconstructing an instrument's bore shape and damping profile from its transient response has been developed. It produces perfect results for noiseless model data, but even the smallest amount of noise renders the method unstable. Regularisation is therefore required. The corresponding direct process of predicting the transient response from bore and damping data is stable and produces results which compare well with measured responses. The work strengthens relationships between an instrument's shape and its musical quality, and will enhance the design of better instruments. Further research on the link between transient response and subjective quality is recommended.