Ryu-ichi Oka

Phoneme labelling of frame by accumulating evidence obtained from spotting results of microscopic units

Comparatively macroscopic phonemic units consisting of continuous sounds are used as standard patterns in continuous dynamic programming. In matching intervals between these standard patterns and input patterns, a sampled pattern is labeled according to the phonemes and phonemic particles which form the standard patterns. In the matching intervals, the labeled pattern is called a matching history. The matching history of a standard pattern when the continuous dynamic programming value is below a certain threshold value is used to recognize the phonemes of the input patterns.

Method of feature determination and extraction and recognition of voice and apparatus therefore

A voice recognition method and apparatus in which a scalar time-space field pattern lattice plane having an abscissa and an ordinate, with one being the time axis and the other the space axis, is obtained from a voice signal inputted as a recognition object, and a voice characteristic pattern based upon said time-space pattern is to be formed as a standard pattern or to be matched with a standard pattern to recognize a characteristic of the inputted voice. The scaler time-space field pattern is converted to extract a vector field pattern lattice corresponding to each lattice point of the scalar time-space field pattern lattice plane with each vector of the vector field pattern lattice having a quantity and an angular orientation value. The angular orientation values of the vector field pattern lattice are placed into one of N groups, each group corresponding to a range of vector angular orientation values, and a set of N two dimensional lattices are formed with each lattice corresponding to one of the vector angular orientation ranges and each point of the lattice having the quantity parameter of the corresponding point of the vector field pattern lattice. The patterns of the N two dimensional lattices have the voice characteristic information can be used as standards or to be matched with standard patterns. There can be blurring processing of the quantity values of the N two dimensional lattices to account for speech variations, such as between male and females, and the resulting lattices of the blurring processing used as standards or for matching.

Speaker-independent word speech recognition using the blurred orientation patterns obtained from the vector field of spectrum

A speech pattern is proposed and used to realize a speaker-independent word recognition system. The algorithm consists of two key concepts: utilization of a novel speech feature called the orientation feature, obtained from the vector field of the spectrum, and the application of a blurring operation to the orientation pattern in order to enhance and stabilize the phonemic features.<<ETX>>

Design and data collection for a spoken dialog database in the Real World Computing (RWC) program

The RWC program is constructing substantial databases for advancing and evaluating research and development conducted under its program and related domains. In this presentation the motivation of this effort, a basic design of spoken dialog databases, and the current status of data collection work are described. At the first stage, some fundamental data collection has been carried out to determine several environmental conditions and data‐filing specifications. Here, two topics are selected for the dialog: one was dialogs between car dealers and customers, and the other was dialogs between travel agents and customers. Professional dealers and agents were employed to produce reality in the conversations. To date, 60 samples of dialogs were recorded and 48 of them were filed into CD‐ROMs which included about 10 h of speech waveforms with transcriptions and labeling‐related information. The speech data are almost completely spontaneous but are of good quality in the acoustic‐phonetic sense. The CD‐ROMs are r...