The self-excitation possibility of underwater organ pipes as the counterpart of aerial ones is considered. Despite of the great differences between water and air, the organ pipe sounds in water as well as in air. The water-jet velocity is formulated by assuming the turbulent jet with weak additional decay. The velocity of the underwater sound propagating in the pipe is theoretically estimated, and the Q-value of the pipe is measured. They depend on the distensibility of the pipe wall material. The margin of selfexcitation is sufficient in aluminium pipes but bare in acryl pipes due to the smallness of Q-value. Numerical calculation well explains the unique experimental excitation characteristics not obtained from aerial organ pipes. Moreover, the overall efficiency of power conversion is estimated.

View full abstract

A unified statistical treatment of the probability distribution function is theoretically proposed in the case when a general random noise of arbitrary distribution type exhibits various type nonstationary properties due to the temporal change of statistical moments. For the purpose of finding the effect of nonstationarity caused by the temporal change of moments on the resultant probability distribution form, the explicit expression of the probability density function is derived in a general form of statistical expansion series taking the stationary term into the first term of series expansion. Next, a new trial of estimating a noise level distribution over a long time interval on the basis of level distribution within a short time interval is discussed, by using the above theoretical probability expression. The validity of the theoretical result is confirmed experimentally by applying it to the actual level data of the nonstationary road traffic noise observed in a large city.

View full abstract

A model for the generation of fundamental frequency contours (F₀ contours) of spoken, sentences is presented for the purpose of elucidating the relationship between the sentence F₀ contour and the linguistic and non-linguistic information. It is based on a quantitative formulation of the process whereby the logarithmic fundamental frequency is controlled in proportion to the sum of two components corresponding respectively to the effects of phrase and accent. The model's parameters were determined to give the best approximation to an observed F₀ contour on the basis of the mean squared error. Analysis of natural utterances of various declarative sentences of Japanese revealedthat the model can generate close approximations to observed F₀ contours from a set of discrete commands and a small number of parameters. The extracted parameters were found to be closely related to linguistic factors and factors constituting thenaturalness of speech. These results provide a means for generating natural F₀ contours from a small set of parameters and rules for synthesis.

View full abstract

This paper reports on the results of the application of a vector quantization technique to an isolated word recognition system. The basic idea underlying this system is to represent a speech spectral sequence by several discrete spectral symbols. In this system, word templates are represented as sequences of discrete phoneme-like (pseudo-phoneme) templates which are automatically generated from a training set of word utterances by a clustering technique. The new word recognition system and its advantages are explained. This recognition system is especially effective for speaker-dependent largevocabulary word recognition, as well as speaker-independent word recognition using multiple word templates.

View full abstract

This paper describes experiments on recognition of vowels in 212 Japanese words. Recognition is based on frequency distribution of spectral local peaks for every frame. The spectrum obtained from a band pass filter bank sampled every 10 ms is represented by a binary vector, each element denoting the presence or absence of a local peak at a particular channel. The conditional probability distribution of local peaks and the joint probability between one local peak and another are obtained by a statistical analysis. Recognition procedures are derived from the functional form of these underlying probability distributions. The power information in certain channels is also used as a supplementary one for improving the recognition accuracies. The recognition rate is 90.8% for designed samples, and 85.3% for test samples uttered by5males. The effects in transfer systems with various frequency characteristics on vowel recognition are investigated. The experimental results indicate that the effects are significant, and the proposed method using local peaks is robust for the different frequency characteristics studied.

View full abstract

A theory of annoyance is developed by supposing a psychological system of a man. Mathematical expressions of the psychological system are derived in a similar way of thermodynamics. An application of the theory to the community response of the bullet train noise is also performed.

View full abstract