以人工類神經網路連結揚聲器客觀量測指標與主觀聽覺屬性之間的相關性

Abstract

揚聲器在大訊號的情形下工作時，許多揚聲器的非線性效應便會顯現，造成聲音的失真進而破壞音質。本研究的目的主要是連結動圈 式揚聲器客觀量測指標與主觀聽覺屬性之間的相關性。利用大訊號揚 聲器模型製造出數個非線性模型，並分別對非線性模型做客觀指標量 測以及主觀聽覺測試。再使用迴歸分析、多變異數分析來分析數據， 並使用Fisher's LSD 檢測主觀聽覺測試結果是否在統計上有顯著的差異。再來我們使用類神經網路來連結客觀量測指標與主觀聽覺屬性之間的相關性。類神經網路對於非線性失真所造成的音質檢測，提供了一個不需要做主觀聽覺測試，而且更有效率的方法。

As a loudspeaker operates in the large-signal domain, nonlinear distortion may arise and impair the sound quality. This work aims to correlate various subjective audio attributes and the objective nonlinear measurements for moving-coil loudspeakers. Several nonlinear models of loudspeaker are created, based on a large-signal loudspeaker model. The data of subjective listening test were processed by the regression analysis, the multivariate analysis of variance (MANOVA), and the least significant difference method (Fisher’s LSD) as a post hoc test to justify the statistical significance of the results. The objective and subjective indices are correlated with the aid of an artificial neural network (ANN). The network proved effective in assessing subjectively the sound quality impairment due to nonlinear distortions of loudspeakers based on only objective measurements, without having to conduct listening tests.