The masked detectability thresholds of amplitude and frequency modulated tones

Work recently carried out at NPL on the assessment of industrial noise has shown the necessity of developing a method to objectively define the magnitude of physical features in noise in terms of the subjective impact of the noise. This work has led to the development of a method to objectively identify the presence of a modulated waveform in background noise by determining its subjective masked detection threshold. We can summarise the three stages of the work in the following: generation of sound stimuli, subjective listening tests and signal analysis. The sound stimuli were either amplitude or frequency modulated tones in a broadband background noise which were generated by means of the MARS (Musical Audio Research Station) workstation and recorded onto a DAT tape. Laboratory listening tests were carried out: subjects were asked to detect the tonal component in order to calculate the mean masked subjective detectability thresholds for each modulation waveform. The signal analysis aimed at developing an objective measure of sensation level in a given band at masked detectability threshold that is not dependent on having prior knowledge of modulation type, frequency and depth by adopting an appropriate integration time for the analysis. The results of the work have led to a measure based on rms maximum levels integrated over 160-180 ms time periods and have implied that this time period may relate to the processing time of the auditory system.