The detrimental effect of reverberation on speech intelligibility is studied on the assumption that speech is series of discreat consonants and vowels. To propose speech model, relative level and average duration of Japanese vowels and consonants are discussed. After being made sure that amplitude probability distribution of speech is similar to that calculated on the proposed model, hearing threshold of consonant was evaluated using simulated signal based on the above speech model. Several articulation tests with artificially reverberated sound were also carried out. These results show that the detrimental effect causes mainly against consonants, however vowel articulation scarcely decreases. Furthermore, calculated percent articulation exhibits good approximation to experimental results. As a conclusion discussed here, it is suggested that the detrimental effect of reverberation depends on both reverberation characteristic and syllable duration time or speech rate. For an example, to attain word articulation above 80%, the reverberation time should be shorter than 0. 75s, where syllable duration time is 100ms.
Introducing acoustic radiation ratio η_r which is ratio of acoustic power to mechanical power of a vibrating plate, frequency dependence of acoustic radiation from circular plates are theoretically studied. η_r has a deep through between the very low eigenfrequencies and the coincidence frequency. When the diameter decreases or the thickness increases, the depth of the though decreases. In general, when the plate becomes thin, the deflection becomes large, and the noise increases. However, in noise reduction, it is important to make use of the through of η_r, because, at some frequency range, a thin plate has smaller η_r than a thick plate.
Statistical analysis were made of static features of voiced sound spectra and threedimensional representation of consonants on the basis of the dynamic features of spectra. The spectral change of consonants, caused by the factor of speaker, vowel and time interval were also investigated and the following significant facts were found from these analysis. (1)Voiced consonant recognition for a limited speaker is easier than vowel recognition for unlimited speakers. (2)The relationship among consonants on the spectral space corresponds to the relationship in articulation. (3)The spectra of /i/, /u/, consonants followed by /i/ or /u/, and nasal are subject to the influence of the speaker-factor more than others for a short time interval between test and reference samples. (4)There is no significant difference of speaker-factor among consonants for a long time interval. (5)Long-term-independent personality in spectrum is extracted more easily from vowel spectrum than others
The attenuation characteristic of the N-fold walls is studied as a function of frequency by taking into consideration the effects of the resistive factor which was firstly introduced by A. London, the coincidence phenomenon and the internal dissipation factor in air space with the aid of the equivalent circuit method(especially introducing an equivalent distributed constant circuit model for the system under consideration)which seems more flexibly applicable than the wave equation method by A. London and the multiple reflection method by K. A. Mulholland. The validity of the present theory is confirmed experimentally using the transmission losses(measured at Kanagawa Prefectural Environmental Center and Hiroshima University)