Audio signal compression method, audio signal compression apparatus, speech signal compression method, speech signal compression apparatus, speech recognition method, and speech recognition apparatus

An audio signal compression method for compressively coding an input audio signal, including the steps of:

• transforming the input signal into a frequency-warped signal with an all-pass filter, using a weighting function on frequency based on human auditory characteristics;
• obtaining a spectrum envelope having different resolutions for different frequencies, by performing linear predictive analysis of the frequency-warped signal;
• and
• flattening the input audio signal for each frame using the spectrum envelope.

An audio signal compression method for compressively coding an input audio signal, said method having the step of performing mel-linear predictive analysis including frequency warping in a prediction model, thereby calculating a spectrum envelope having different resolutions for different frequencies, from the input audio signal, using a weighting function on frequency based on human auditory characteristics;

• and
• said mel-linear predictive analysis comprising the steps of:
• cutting out an input signal of a specific time length from the input audio signal, and filtering the signal of the time length using multiple stages of all-pass filters to obtain output signals from the respective filters;
• obtaining an autocorrelation function on a mel-frequency axis by performing a product-sum operation between the input signal and the output signal from each filter, which product-sum operation is performed within a range restricted to the time length of the input signal as represented by the following formula, φ ⁢ ( i , j ) = ∑ n = 0 N - 1 ⁢ x ⁡ [ n ] · y ( i - j ) ⁡ [ n ]
• wherein φ(i,j) is the autocorrelation function, x[n] is the input signal, and y(i−j)[n] is the output signal from each filter;
• obtaining mel-linear predictive coefficients from the autocorrelation function on the mel-frequency axis;
• and
• using the mel-linear predictive coefficients as a spectrum envelope, or obtaining a spectrum envelope from the mel-linear predictive coefficients.

An audio signal compression apparatus for compressively coding an input audio signal, comprising:

• time-to-frequency transformation means for transforming the input audio signal to a frequency domain signal;
• spectrum envelope calculation means for calculating a spectrum envelope having different resolutions for different frequencies, from the input audio signal, using a weighting function on frequency based on human auditory characteristics;
• normalization means for normalizing the frequency domain signal using the spectrum envelope to obtain a residual signal;
• power normalization means for normalizing the residual signal by the power;
• auditory weighting calculation means for calculating weighting coefficients on frequency, based on the spectrum of the input audio signal and human auditory characteristics;
• and
• multi-stage quantization means having plural stages of vector quantizers connected in series, to which the normalized residual signal is input, and at least one of the vector quantizers quantizing the residual signal using the weighting coefficients.

An audio signal compression apparatus for compressively coding an input audio signal, comprising:

• mel-parameter calculation means for calculating mel-linear predictive coefficients on a mel-frequency axis which represents a spectrum envelope having different resolutions for different frequencies, from the input audio signal, using a weighting function on frequency based on human auditory characteristics;
• parameter transformation means for transforming the mel-linear predictive coefficients to parameters representing a spectrum envelope, such as linear predictive coefficients on a linear frequency axis;
• envelope normalization means for normalizing the input audio signal by inversely filtering it with the parameters representing the spectrum envelope, to obtain a residual signal;
• power normalization means for normalizing the residual signal using the maximum value or mean value of the power to obtain a normalized residual signal;
• and
• vector quantization means for vector-quantizing the normalized residual signal using a residual code book to transform the residual signal into residual codes.

A speech signal compression method for compressively coding an input speech signal, said method having the step of performing mel-linear predictive analysis including frequency warping in a prediction model, thereby calculating a spectrum envelope having different resolutions for different frequencies, from the input speech signal, using a weighting function on frequency based on human auditory characteristics;

• and
• said mel-linear predictive analysis comprising the steps of:
• cutting out an input signal of a specific time length from the input speech signal, and filtering the signal of the time length using multiple stages of all-pass filters to obtain output signals from the respective filters;
• obtaining an autocorrelation function on a mel-frequency axis by performing a product-sum operation between the input signal and the output signal from each filter, which product-sum operation is performed within a range restricted to the time length of the input signal as represented by the following formula, φ ⁢ ( i , j ) = ∑ n = 0 N - 1 ⁢ x ⁡ [ n ] · y ( i - j ) ⁡ [ n ]
• wherein φ(i,j) is the autocorrelation function, x [n] is the input signal, and y(i−j)[n] is the output signal from each filter;
• obtaining mel-linear predictive coefficients from the autocorrelation function on the mel-frequency axis;
• and
• using the mel-linear predictive coefficients as a spectrum envelope, or obtaining a spectrum envelope from the mel-linear predictive coefficients.

A speech signal compression apparatus for compressively coding an input audio signal, comprising:

• mel-parameter calculation means for calculating mel-linear predictive coefficients on a mel-frequency axis which represents a spectrum envelope having different resolutions for different frequencies, from the input speech signal, using a weighting function on frequency based on human auditory characteristics;
• parameter transformation means for transforming the mel-linear predictive coefficients to parameters representing a spectrum envelope, such as linear predictive coefficients on a linear frequency axis;
• envelope normalization means for normalizing the input signal by inversely filtering it with the parameters representing the spectrum envelope, to obtain a residual signal;
• power normalization means for normalizing the residual signal using the maximum value or mean value of the power to obtain a normalized residual signal;
• and
• vector quantization means for vector-quantizing the normalized residual signal using a residual code book to transform the residual signal into residual codes.