Abstract: A method for estimating the probability of phone boundaries and the accuracy of the acoustic modelling in reducing a search-space in a speech recognition system. The accuracy of the acoustic modelling is quantified by the rank of the correct phone. The system includes a microphone for converting an utterance into an electrical signal, which is processed by an acoustic processor and label match which finds the best-matched acoustic label prototype. A probability distribution on phone boundaries is produced for every time frame using a first decision tree. These probabilities are compared to a threshold and some time frames are identified as boundaries between phones. An acoustic score is computed for all phones between every given pair of hypothesized boundaries, and the phones are ranked on the basis of this score. A second decision tree is traversed for every time frame to obtain the worst case rank of the correct phone at that time, and a short list of allowed phones is made for every time frame.

Abstract: Respective samples of input speech data are transformed by a discrete Fourier transformer to obtain a spectrum of the speech data. Simultaneously, values of times of the respective samples are multiplied with the input speech data by a multiplier and a differential spectrum is obtained by transforming a result of the multiplication by a discrete Fourier transformer. A real part of a value obtained by dividing the differential spectrum by the spectrum by a quotient real part calculator and the real part is inverse-transformed by a discrete inverse Fourier transformer. The result of the inverse-transformation is divided by the values of the times of the respective samples to obtain a time function corresponding to phase. On the other hand, a time function corresponding to a logarithmic amplitude spectrum is obtained from an output of the inverse-Fourier transformer by means of a logarithmic amplitude spectrum calculator.

Abstract: A speech decoder apparatus for synthesizing a speech signal from a digitized speech bit stream of the type produced by processing speech with a speech encoder. The apparatus includes an analyzer for processing the digitized speech bit stream to generate an angular frequency and magnitude for each of a plurality of sinusoidal components representing the speech processed by the speech encoder, the analyzer generating the angular frequencies and magnitudes over a sequence of times; a random signal generator for generating a time sequence of random phase components; a phase synthesizer for generating a time sequence of synthesized phases for at least some of the sinusoidal components, the synthesized phases being generated from the angular frequencies and random phase components; and a synthesizer for synthesizing speech from the time sequences of angular frequencies, magnitudes, and synthesized phases.

Abstract: A method for determining the block size of a transform coder in which the digital audio signals having spectral and temporal structure are decomposed into plural spectral frames. The method involves defining the audio signal into time intervals according to the temporal masking properties of a human auditory system; obtaining a peak value in each of these time intervals; calculating differences among the peak values; and selecting a block size based on a comparison of the differences against a predefined value.

Abstract: A method of estimating the pitch of a speech acoustic signal in a time interval in which said signal is a voiced one, wherein the pitch corresponds to the distance between the contact points of a circle and a plot, normalized to a limit value, of the energy of said speech acoustic signal as a function of time; said contact points being obtained by rolling said circle on said plot.

Abstract: A set of intonation intervals for a chosen dialect are applied to the intonational contour of a phomene string derived from a single set of stored linguistic units, e.g., phonemes. Sets of intonational intervals are stored to simulate or recognize different dialects or languages from a single set of stored phonemes. The interval rules preferably use a prosodic analysis of the phoneme string or other cues to apply a given interval to the phoneme string. A second set of interval data is provided for semantic information. The speech system is based on the observation that each dialect and language possess its own set of musical relationships or intonation intervals. These musical relationships are used by a human listener to identify the particular dialect or language. The speech system may be either a speech synthesis or speech analysis tool or may be a combined speech synthesis/analysis system.

Abstract: An improved text-to-speech synthesizer that employs a text to speech converter, a text reader control procedure, a classifier procedure, an abbreviation expansion procedure, and an acronym/initialism expanding procedure is herein described. A classifier procedure is used to classify generate classification values for each word in the text message with regard to syntax, punctuation and membership in predefined classes of words, the predefined classes of words including number, measurement units, geographic designations, and date/time values. An abbreviation expansion procedure evaluates, based on the classification values for words neighboring the identified words, which, if any, of the potential expansion values is applicable, and substitutes the potential expansion for the identified abbreviation word when evaluation yields a success value.