A previous study has demonstrated that speech intelligibility is improved for a tone language when sine-wave speech is noise-vocoded, because noise-vocoding eliminates the quasi-periodicity of sine-wave speech. This study examined whether identification accuracy of Japanese pitch-accent words increases after sine-wave speech is noise-vocoded. The results showed that the Japanese listeners’ identification accuracy significantly increased, but their discrimination accuracy did not show a significant difference between the sine-wave speech and noise-vocoded sine-wave speech conditions. These results suggest that Japanese listeners can auditorily discriminate minimal-pair words using any acoustic cues in both conditions, but quasi-periodicity is eliminated by noise-vocoding so that the Japanese listeners’ identification accuracy increases in the noise-vocoded sine-wave speech condition. The same results were not observed when another way of noise-vocoding was used in a previous study, suggesting that the quasi-periodicity of sine-wave speech needs to be adequately eliminated by a noise-vocoder to show a significant difference in identification.
Read full abstract