Development of a modulation transfer function-based method for evaluating bass reproduction accuracy in professional monitoring loudspeakers

This study develops a measure which allows visual and quantitative assessment of bass reproduction accuracy in professional studio monitors. This type of loudspeaker must present mix engineers in particular with a faithful impression of recordings; they can then create an optimum musical balance between instruments that will translate well to other reproduction systems. Inaccurate monitors can lead to expensive remixing or a degraded musical experience for the end consumers, especially if the fundamental rhythm section has been poorly adjusted. Existing work suggested that the Modulation Transfer Function (MTF) might be a more informative descriptor of bass reproduction accuracy than typical steady-state measures; it might therefore provide a grading system of performance that engineers could use when selecting suitable monitors for their work. The purpose of this project was to investigate the technique and i) develop an algorithm to summarise the critical aspects of monitor performance at low frequencies and ii) see whether it predicted the subjective impression of reproduction accuracy. An algorithm was developed, considering different calculation methods and parameters to optimise it for low-frequency application with musical signals. It was applied to groups of loudspeaker models, simulating the responses of real mix monitors; then listening tests were conducted with these models reproducing music. The subjective and objective results were compared to see whether the algorithm would be a useful measure of monitor performance. The algorithm successfully summarised behaviour of simulated and measured monitor responses; it described important factors such as extension and smoothness, and how the system responded to temporally-varying input signals. Based on ordinal data, the algorithm was found to predict all statistically significant judgements from listeners. These participants had demonstrated that they were accurate and consistent listeners, but found it difficult to reach consensus in some evaluations where the listening task required more skilled judgements of overall performance. It was concluded that the algorithm in its current form is effective and suitable for the intended application, but subjective evaluations of more complex alignments are needed from professional mix engineers; this might allow the numerical MTF scores to be graded against perceived bass reproduction accuracy, therefore enhancing the predictive power of the technique.