Two-tone masking and auditory critical bandwidths.

Two-tone masking of 1,000-Hz signals was investigated in 5 normally-hearing listerners. The results indicate a consistent increase in critical bandwidth with level of the two masking tones (approximately 3%/dB). Irregularities in the individual two-tone masking functions led to further investigations with a single listener. For subcritical frequency separations between masking tones, -10 dB/decade masking slopes were demonstrated along with large dips or notches in the two-tone masking functions. Both the -10 dB/decade masking slopes and the large notches were replicated in detail when only the lower-frequency tones of the two-tone masking pairs were employed as maskers. Separate masking bands were then introduced that eliminated the -10 dB/decade masking slope and the large notch. It is concluded that the two-tone masking experiment is really a single-tone experiment, and that the detection of beats, roughness, and combination tones determines the form of single-tone and two-tone masking functions. If critical bnadwidth estimates are to be obtained from these experiments, it is most appropriate to refer to those estimates as "critical bandwidths for tonal interaction".