This paper describes a speaker independent word recognition algorithm that is based on four layer neural networks with embedded eigenvectors. Eigenvectors from the sub-space method (SM) are used as weights for the first hidden layer. A similarity measure given by SM is calculated by cumulative summation of the projection components of aninput pattern onto a set of eigenvectors. In contrast to this, our new method evaluates each projection component to achieve better performance than SM. We propose the subspace training (SST) algorithm with SM and the decision controlled back propaga-tion training (DCBPT) algorithm to improve recognition performance and to reduce training times. Training and recognition experiments were performed using a 26 word vocabulary consisting of train station names. The error rate was 1.3% using SM and was reduced to 0.7% using the combination of neural networks and SM.

抄録全体を表示

The piezoelectric vibrator has been used at its resonance in general; therefore the equiv-alent circuits have been obtained as an admittance-type expression drawn from the input admittance developed near their resonance point. However, it may not neces-sarily be well for some piezoelectric devices, because they cannot yield a superior char-acteristic when used at the resonance frequency. On the other hand, vibrational char-acteristics of a piezoelectric ceramics vibrator only are proved experimentally to be good at its anti-resonance point. Therefore, the authors thoughthat the equivalent circuit of a piezoelectric vibrator should be expressed depending upon its anti-resonance, where it is called the B-type equivalent circuit. This paper deals with the equivalent circuits developed near the anti-resonance mainly depending upon the input impedance of the piezoelectric vibrator.

抄録全体を表示

Precision calibration of echo sounder by integration of sphere echoes is studied. In this method, the total sensitivity of the system is easily and quickly determined by comparing the echo integrator output for standard sphere echoes with the theoretical value. A strict theoretical expression is derived considering the bandwidth of receiving system and the time varied gain (TVG) effect on the sphere echo shape. The strict expression is compared with approximate ones and with experimental values. Experimental values agree fairly well with the theoretical calculations. For precise calibration, the effective pulse width should be calculated from the actual echo shape since the bandwidth of the receiving system changes the echo shape. When a long pulse width is used, the TVG effect should not be neglected.

抄録全体を表示

This paper deals with the development and the validation of a model used to evaluate the performance of parallel noise barrier configuration. These barriers are generally used to reduce the noise emitted by traffic roads. The formulation and algorithm used combine the multiple reflections occuring between the parallel barriers by the acoustical imagery method, and the barriers absorption by the modified Delany-Bazley impedance model. They also take into account parallel barriers with different heights, the ground effect, and a spectral sound power for various type of traffic noise. A complex summation between acoustical waves originating from a single source is used to obtain rigorously constructive and destructive acoustical interferences. This approach has been validated against systematic experimentations in controlled conditions in a semi-anechoic room. A case study is also presented which highlights the importance of applying absorbent materials on the barriers in terms of an increase in efficiency.

抄録全体を表示

Conventional pulsed ultrasound Doppler technique has a fundamental limitation in the product of maximum velocity and maximum range because the Doppler shifted fre-quency is evaluated from the pulsed Doppler signal sampled at pulse repetition fre-quency, and the range is also restricted by pulse repetition frequency. This limitation sometimes causes the confusion in the measurement of blood flow velocity in a human heart. This paper describes a principle of alias-free interpolation technique for pulsed Doppler signal. This technique enables to synthesize the Doppler signal with high Nyquist frequency by interpolating the roughly spaced samples of the Doppler signal. The velocity range is expanded, and minimum sampling interval is not less than that in conventional technique, so that the range-velocity product can be expanded with this technique. Two cases of the interpolation are shown here. One is the interpolation for the single component Doppler signal and the other is for the two component Doppler signal. It is shown in simulation that these interpolation make the range-velocity product double.

抄録全体を表示

The LVQ-HMM hybrid algorithm was one of the first algorithms proposed in a recent approach aiming to integrate a highly discriminative artificial neural network-based classifier with an HMM capable of representing temporal structure effectively. The high phoneme classification capability of LVQ-HMM has already been demonstrated. However, the performance of LVQ-HMM has been less striking in more difficult, large-scale speech recognition situations, making evaluation of the algorithm controversial and suggesting a more detailed investigation of the properties of the algorithm in such situations. This technical report is thus devoted to re-evaluation of the hybrid algo-rithm, evaluated for word and phrase recognition tasks. Specifically, recognition ex perimentsare conducted under rather difficult, speaker-independent and large-vocabu-lary conditions. Our recognizer uses a phoneme-based strategy; in particular, the predictive LR-parser is incorporated for efficient recognition. Experimental results alone are unfortunately insufficient to cease the controversy. However, possible contribu-tions and aspects of the algorithm needing further improvement are brought to light.

抄録全体を表示

A new equipment is developed to supply very fine powder using ultrasonic vibration. It has been well known that very fine powder inside a hopper forms easily powder-bridge. This behavior prevents the powder from continuously flowing. The author, however, utilizes the behavior in this study. Namely, the powder on a plate with multiholes of a suitable diameter can not flow out through the holes because of the powder-bridge, when the plate is not vibrating. On the other hand, the powder can flow out continuously, when the plate begins vibrating. The flow rate can be easily controlled by holding vibrational velocity of the plate at a fixed level. The paper deals with fundamental studies for this kind of powder supply.

抄録全体を表示