Spectral factors involved in the intelligibility of speech
MetadataShow full item record
Monosyllabic word intelligibility was examined under a variety of conditions. First/ white and pink noise were used to mask speech. Pink noise was found to mask better than white noise for all values of signal-to-noise ratio. Computations of the intelligibility of speech in white and pink noise spectra, using the 20-band equal articulation method, were found to agree with the experimental results. Monosyllabic word intelligibility was measured for speech that was filter-enhanced with a center frequency of 2 kHz. The results here suggest that those enhancements result in significant changes in voice quality, but do not noticeably affect speech intelligibility. These results were also consistent with computations based upon the 20- band articulation index model. Lastly, monosyllabic word intelligibility was measured for two voice modes produced by a professional stage actor. The two speech modes, called the normal and stage voice modes, were found to differ in their vowel formant patterns. The stage voice vowel sounds have lower second and third formant frequencies and sound darker and richer in tone. Speech intelligibility was measured for the two voice modes. The results in this experiment suggest that the use of the stage voice does not result in any significant change in the speech intelligibility.