Self-masking effects from live and recorded vowels.
As a means of determining the masking characteristics of spoken (live) vowels, as compared to recorded vowels, 16 normal Ss were trained to track pure-tone swept-frequency thresholds in the presence of vowels used as masking noise. While tracking thresholds, Ss either phonated the vowels /a/, /i/, or /u/ continuously or listened to a taped playback of their own voices. The resulting continuous-frequency audiograms generally resembles the configurations of the vowel spectra. Overall masking effect was significantly greater for the recorded vowels (air-conducted feedback only) than for the live vowels (both air- and bone-conducted feedback) at equal loudness. Recorded vowels tended to have greater masking effect in the higher frequencies, while live vowels tended to have greater effect in the lower frequencies.