| Abstrakt: |
Speech of 4 adult males with general American dialect, but with considerably different vocal gesture, was recorded and filtered such that at the widest bandwidth (1500 cps) and optimum center frequency (1600 cps) intelligibility of key words in our revisions of the CHABA medico-legal sentences remained high (95%). Untrained listeners were used, who never heard the same sentence twice. Further distortions of shouting, interruption (8/sec, 0.5 duty cycle), and reverberation (about 3 sec through 1 kc) lowered the intelligibility slightly (12% reduction by shouting) or very markedly (80% reduction by reverberation). For the narrowest bandwidth used (500 cps), the optimum center frequency for all distorted speech was shifted upward to 1900 cps or above. Even with the lowest center frequency used (800 cps), intelligibility remained nearly as good as at the next frequency up (1200 cps), at 1900 cps. The principle of frequency equipotentiality is operating here. The articulation index as it relates to distorted speech is discussed, and modifications are suggested. [ABSTRACT FROM AUTHOR] |
