An algorithm is proposed in this paper for phonemic segmentation to improve the performance of a continuous Mandarin speech recognition systems. The coefficient of time variation of spectral envelope and the coefficient of tirrie. variation of zero order cepstrum are extracted using Unbiased Estimation of Log Spectrum (UELS). The parameter curves based on these coefficients are very smooth, therefore, the relation between parameter's maximum values and phoneme boundaries are easy to be found. By these smooth curves, the maximum value can be used as a criterion to delimit phonemes, rather than the threshold that is used in conventional systems, hence it is possible to get precise segmentation results. 300 sentences were used for an experiment, and the results show the system performance is better than traditional methods. The average phoneme-deletion rate is 1.3%, average phonemeinsertion rate is 3%. For evaluation, the segmentation results were used for a phoneme recognition experiment. 95.5% consonants recognition rate and 92.5% vowel recognition rate were obtained. The results show the approach is highly effective.

抄録全体を表示

The loudness of a 1, 600-Hz pure-tone signal partially masked by a 1, 000-Hz, 70-dB SPL masker is experimentally examined using the loudness matching method. As for two of the four subjects who participated in the experiment, the loudness of the signal for signal levels higher than 45 dB SPL was found to be enhanced in spite of the masking. This phenomenon disappeared in the presence of one-third-octave band noise whose center frequency was 2, 500 Hz. These results suggest that judgments of the loudness are affected by aural combination tones when the masker used is a pure tone. Five loudness functions proposed earlier were tested to determine if they can explain the loudness of a pure tone partially masked by another pure tone. As a result, the formula using the driven firing rate function of the auditory sensory receptor shows good agreement with some of the data.

抄録全体を表示

“Vocalness”, which indicates correctness of vocalization of singing voice used in western classical opera and lied, is defined in terms of relative intensity of the singing formant. An advantage of vocalness is that its measurement is scarcely affected by the vibrato, inherent in the singing voice. As a result of vocalness measurement for sustained vowel /ah/ sung by 21 female singers, no significant difference was found between the two groups of (A) professional singers, and (B) students of voice and trained chorus singers, whereas a marked difference was found between these groups and the group of beginning amateur chorus singers (C). Thus vocalness is expected to be a helpful measure for the vocalization training of beginners.

抄録全体を表示

We conducted studies regarding sensitivity changes of Electret Condenser Microphones (ECMs) that were stored in a laboratory for 18 years, and using the results, estimated the working lifetime of an ECM. The average sensitivity change of the ECMs stored at room temperature for 18 years was unexpectedly small-only about 0.3dB. We estimated the main factor that determines the lifetime of an ECM-surface charge decreases of electret foil-over a period of time, using a heat acceleration test and the results of our long-term sensitivity change studies. The results of our estimate showed that a sensitivity decrease of 3dB in ECMs stored at room temperature may be found in 60 years, and a 6 dB decrease, which is the limit of microphone performance, may be found after 200 years. In this paper, we propose a form of expression for the decrease pattern of electret surface charges, over a period of time, and temperature as a three dimensional form.

抄録全体を表示

The transmission property of the fundamental and second harmonic components of the focused Gaussian beam incident on a solid plate put perpendicular to the acoustic axis has been investigated. In the present theory, the dependence of the transmission coefficient of plane wave on the incidence angle is applied to the fundamental and second harmonic pressures of the incident focusing beam, which has been decomposed in plane waves using Hankel transform. After the passage of the sound through the plate, the ensemble of plane waves is reconstructed to a beam, by means of the inverse Hankel transform. The nonlinear property which might exist in solid plates is neglected. The nonlinear propagation in the liquid is described by the Khokhlov-Zabolotskaya-Kuznetsov equation, which is solved under the successive approximation. The experiment for the 1.9-MHz focused sound transmitting through a silica glass or polystyrene plate immersed in water, and a methanol layer between two polystyrene plates, validates the present theory.

抄録全体を表示