Study on tone-to-noise ratio calculation method for tones with frequency and amplitude variation

This paper is to present a method on improving consistency and accuracy of Tone-to-Noise Ratio (TNR) calculation for varying frequency tones in reporting prominent discrete tone according to ECMA-418-1. For given sound pressure spectrum, existing TNR calculation methods require suspect tones to be selected by the analyst, who must also determine the width of the tone, that could introduce significant uncertainty and variation between different analysts. Moreover, tones associated with unstable frequency could cause even more uncertainty in determining tone frequency band. In this study, an approach is proposed to determine the existence of discrete tone, define the tone frequency band, and lastly calculate TNR such that different analysts will be able to obtain consistent and more accurate TNR value. To validate the approach, several different noise spectra were used as mask noise, and thousands of numerically generated tones, with randomly varying frequencies and amplitudes, were modulated to the mask noise to generate synthesized waveform for TNR calculation. These randomly generated synthesized waveform were subjectively evaluated and confirmed that prominent discrete tone criteria are still applicable to tones with varying frequencies and amplitudes. The results show that proposed approach can significantly improve TNR calculation accuracy while ensure consistency between different analysts.