A speech analyzer for extracting spectrum information and pitch information from natural speech wherein an accuracy of pitch extraction is enhanced by sampling pitch at a sampling frequency which is higher than a sampling frequency for analyzing the spectrum information.

Storage space in this speech synthesizer is minimized by features including multiplex storage of voiceless phonemes and use of a word designator which selects in parallel information from a word memory and phoneme memory.

Samples of a transmit path voice channel signal are averaged and compared with an adaptive speech threshold to determine the presence of speech. When the average falls below a fixed threshold, a timing circuit is triggered to time a delay period followed by a noise averaging period. The timing of these periods is aborted if either the transmit path average, or a similarly produced receive path average, exceeds the fixed threshold during either period. During the averaging period the voice channe...

This PARCOR-type speech synthesizer replaces a ten-stage lattice type filter with a pipeline multiplier and feedback loop, and provides a loss circuit (for bandwidth broadening) using subtraction circuits for multiplication.

In a speech prosthesis, a direct radiator speaker, a rubber membrane over a speaker for sealing against liquid and food, and an O-ring seal for a battery compartment.

In a speech recognizer, for recognizing unknown utterances in isolated-word speech or continuous speech, improved recognition accuracy is obtained by augmenting the usual spectral representation of the unknown utterance with a dynamic component. A corresponding dynamic component is provided in the templates with which the spectral representation of the utterance is compared. In preferred embodiments, the representation is mel-based cepstral and the dynamic components comprise vector differences ...

Speech recognition using self-correlation (auto-correlation) parameters and time normalization is improved by first making a preliminary decision using lower-order (zeroth to third) self-correlation coefficients derived from the self-correlation function of order m: ##EQU1## where .phi.(0) is the power, .phi.(1) is the linear or first order coefficient, and so on.

In a speech synthesizer, speech speed is selectable by a switch which selects the number of repetitions of a basic waveform or equivalent parameters thereof.

In a speech processor such as a speech recognizer, the problem of distortion of extracted features caused by adaptation of the input automatic gain control (AGC) during feature extraction is solved by storing the AGC's gain coefficient along with the energy level of each extracted feature. At the end of the sampling period the stored gain coefficients are set equal to the minimum stored coefficient and the associated energy levels adjusted accordingly. The AGC circuit may comprise a digitally sw...

Coded text is converted to phonetic data to drive a synthesis filter. Accent data are also obtained to derive a pitch contour for a variable pitch excitation source. Recognition of the beginning of a paragraph causes a pitch contour of higher pitch than the pitch at a later part of the paragraph. The initial pitch falls following each subgroup into which phrases are divided. Accents within a phrase are assigned pitch values which are high for the first accent, less high for the last; and the rem...