Apparatus and method of coding and decoding vocal sound data based on phoneme

Claim 1

An apparatus for coding vocal sound data signals, comprising: transforming means for orthogonal-transforming sequential input digital vocal sound data signals per predetermined number of the data signals to obtain power spectrum data signals; converting means for converting the power spectrum data signals into a data conversion form that features corresponding to phonemes of the vocal sound data signals are extracted; comparing means for comparing converted data signals from the converting means with reference data pattern signals related to the features corresponding to the data conversion form to obtain correlation data signals between the converted data signals and the reference data pattern signals; first extracting means for extracting pitch signals in a frequency direction on the basis of the power spectrum data signals or the converted data signals; second extracting means for extracting power value signals on the basis of the vocal sound data signals or the power spectrum data signals; and coding means for sequentially coding the correlation data signals, pitch signals, and power value signals, and outputting sequential coded vocal sound data signals.

Claim 2

An apparatus for coding vocal sound data signals, comprising: transforming means for orthogonal-transforming sequential input digital vocal sound data signals per predetermined number of the data signals to obtain power spectrum data signals; converting means for converting the power spectrum data signals into a data conversion form that features corresponding to phonemes of the vocal sound data signals are extracted; comparing means for comparing converted data signals from the converting means with reference data pattern signals related to the features corresponding to the data conversion form to obtain correlation data signals between the converted data signals and the reference data pattern signals; first extracting means for extracting pitch signals in a frequency direction on the basis of the power spectrum data signals or the converted data signals; second extracting means for extracting power value signals on the basis of the vocal sound data signals or the power spectrum data signals; and coding means for sequentially coding the correlation data signals, pitch signals, and power value signals, and outputting sequential coded vocal sound data signals.

Claim 6

An apparatus for decoding vocal sound data signals, comprising: decoding means for decoding coded data signals formed at least with pitch signals, and power value signals per phoneme related to vocal sound, the coded data signals being formed by orthogonal-transforming sequential digital vocal sound data signals per predetermined number of the data signals to obtain power spectrum data signals, converting the power spectrum data signals into a data conversion form that features corresponding to phonemes of the vocal sound data signals are extracted, comparing the converted data signals with reference data pattern signals related to the features corresponding to the data conversion form to obtain correlation data signals between the converted data signals and the reference data pattern signals, extracting the pitch signals in a frequency direction on the basis of the power spectrum data signals or the converted data signals, extracting the power value signals on the basis of the vocal sound data signals or the power spectrum data signals and sequentially coding the correlation data signals, pitch signals, and power value signals; signal forming means for forming signals related to each phoneme on the basis of the power value signals and pitch signals decoded by the decoding means; and synthesizing means for synthesizing the formed signals with each other to reproduce vocal sound signals.

Claim 7

An apparatus for decoding vocal sound data signals, comprising: decoding means for decoding coded data signals formed at least with pitch signals, and power value signals per phoneme related to vocal sound, the coded data signals being formed by orthogonal-transforming sequential digital vocal sound data signals per predetermined number of the data signals to obtain power spectrum data signals, converting the power spectrum data signals into a data conversion form that features corresponding to phonemes of the vocal sound data signals are extracted, comparing the converted data signals with reference data pattern signals related to the features corresponding to the data conversion form to obtain correlation data signals between the converted data signals and the reference data pattern signals, extracting the pitch signals in a frequency direction on the basis of the power spectrum data signals or the converted data signals, extracting the power value signals on the basis of the vocal sound data signals or the power spectrum data signals and sequentially coding the correlation data signals, pitch signals, and power value signals; signal forming means for forming signals related to each phoneme on the basis of the power value signals and pitch signals decoded by the decoding means; and synthesizing means for synthesizing the formed signals with each other to reproduce vocal sound signals.

Claim 8

A method for coding vocal sound data signals, comprising the steps of: orthogonal-transforming sequential input digital vocal sound data signals per predetermined number of the data signals to obtain power spectrum data signals; converting the power spectrum data signals into a data conversion form that features corresponding to phonemes of the vocal sound data signals are extracted; comparing converted data signals thus converted into the data conversion form with reference data pattern signals related to the features corresponding to the data conversion form to obtain correlation data signals between the converted data signals and the reference data pattern signals; extracting pitch signals in a frequency direction on the basis of the power spectrum data signals or the converted data signals; extracting power value signals on the basis of the vocal sound data signals or the power spectrum data signals; and sequentially coding the correlation data signals, pitch signals, and power value signals, and outputting sequential coded vocal sound data signals.

Claim 9

A method for coding vocal sound data signals, comprising the steps of: orthogonal-transforming sequential input digital vocal sound data signals per predetermined number of the data signals to obtain power spectrum data signals; converting the power spectrum data signals into a data conversion form that features corresponding to phonemes of the vocal sound data signals are extracted; comparing converted data signals thus converted into the data conversion form with reference data pattern signals related to the features corresponding to the data conversion form to obtain correlation data signals between the converted data signals and the reference data pattern signals; extracting pitch signals in a frequency direction on the basis of the power spectrum data signals or the converted data signals; extracting power value signals on the basis of the vocal sound data signals or the power spectrum data signals; and sequentially coding the correlation data signals, pitch signals, and power value signals, and outputting sequential coded vocal sound data signals.

Claim 15

A method for decoding vocal sound data signals, comprising the steps of: decoding coded data signals formed at least with pitch signals, and power value signals per phoneme related to vocal sound, the coded data signals being formed by orthogonal-transforming sequential digital vocal sound data signals per predetermined number of the data signals to obtain power spectrum data signals, converting the power spectrum data signals into a data conversion form that features corresponding to phonemes of the vocal sound data signals are extracted, comparing the converted data signals with reference data pattern signals related to the features corresponding to the data conversion form to obtain correlation data signals between the converted data signals and the reference data pattern signals, extracting the pitch signals in a frequency direction on the basis of the power spectrum data signals or the converted data signals, extracting the power value signals on the basis of the vocal sound data signals or the power spectrum data signals and sequentially coding the correlation data signals, pitch signals, and power value signals; forming signals related to each phoneme on the basis of the power value signals and pitch signals decoded by the decoding step; and synthesizing the formed signals with each other to reproduce vocal sound signals.

Claim 16

A method for decoding vocal sound data signals, comprising the steps of: decoding coded data signals formed at least with pitch signals, and power value signals per phoneme related to vocal sound, the coded data signals being formed by orthogonal-transforming sequential digital vocal sound data signals per predetermined number of the data signals to obtain power spectrum data signals, converting the power spectrum data signals into a data conversion form that features corresponding to phonemes of the vocal sound data signals are extracted, comparing the converted data signals with reference data pattern signals related to the features corresponding to the data conversion form to obtain correlation data signals between the converted data signals and the reference data pattern signals, extracting the pitch signals in a frequency direction on the basis of the power spectrum data signals or the converted data signals, extracting the power value signals on the basis of the vocal sound data signals or the power spectrum data signals and sequentially coding the correlation data signals, pitch signals, and power value signals; forming signals related to each phoneme on the basis of the power value signals and pitch signals decoded by the decoding step; and synthesizing the formed signals with each other to reproduce vocal sound signals.