Acoustic signal processing apparatus and method, signal recording apparatus and method and program

An acoustic signal processing method for detecting a featuring portion in acoustic signals, comprising:

• an amplitude calculating step of calculating short-term amplitudes, at an interval of a preset time length in the acoustic signals;
• a candidate domain extraction step of extracting candidate domains of said featuring portion based on said short-term amplitudes;
• a feature extraction step of extracting sound quality featuring quantities, quantifying the sound quality, at an interval of said preset time length in the acoustic signals;
• and
• a candidate domain evaluating step of evaluating, based on said short-term amplitudes and said sound quality featuring quantities, whether or not said candidate domain is said featuring portion.

A program for having a computer execute the processing of detecting a featuring portion in acoustic signals, said program comprising:

• an amplitude calculating step of calculating short-term amplitudes, at an interval of a preset time length in the acoustic signals;
• a candidate domain extraction step of extracting candidate domains of said featuring portion based on said short-term amplitudes;
• a feature extraction step of extracting sound quality featuring quantities, quantifying the sound quality, at an interval of said preset time length in the acoustic signals;
• and
• a candidate domain evaluating step of evaluating, based on said short-term amplitudes and said sound quality featuring quantities, whether or not said candidate domain is said featuring portion.

A signal recording apparatus comprising:

• amplitude calculating means for calculating short-term amplitudes, at an interval of a preset time length in acoustic signals;
• candidate domain extraction means for extracting candidate domains of crucial portions of said acoustic signals based on said short-term amplitudes;
• feature extraction means for extracting sound quality featuring quantities, quantifying the sound quality, at an interval of said preset time length in the acoustic signals;
• candidate domain evaluating means for calculating the crucialness of sad candidate domain based on said short-term amplitudes and said sound quality featuring quantities, and for evaluating, based on said short-term amplitudes and said sound quality featuring quantities, whether or not said candidate domain is said featuring portion;
• index generating means for generating the index information including at least a start position and an end position of the candidate domain and the degree of crucialness of the candidate domain evaluated as being said crucial portion by said candidate domain evaluating means;
• and
• recording means for recording said index information along with said acoustic signals.

A signal recording method comprising:

• an amplitude calculating step of calculating short-term amplitudes, at an interval of a preset time length in acoustic signals;
• a candidate domain extraction step of extracting candidate domains of crucial portion of said acoustic signals, based on said short-term amplitudes;
• a feature extraction step of extracting sound quality featuring quantities, quantifying the sound quality, at an interval of said preset time length in the acoustic signals;
• a candidate domain evaluating step of calculating the crucialness of sad candidate domain based on said short-term amplitudes and said sound quality featuring quantities, and for evaluating, based on said short-term amplitudes and said sound quality featuring quantities, whether or not said candidate domain is said featuring portion;
• an index generating step of generating the index information including at least a start position and an end position of the candidate domain and the degree of crucialness of the candidate domain evaluated as being said crucial portion by said candidate domain evaluating means;
• and
• a recording step of recording said index information along with said acoustic signals.

A program for having a computer execute a preset processing, said program comprising:

• an amplitude calculating step of calculating short-term amplitudes, at an interval of a preset time length in acoustic signals;
• a candidate domain extraction step of extracting candidate domains of a crucial portion of said acoustic signals based on said short-term amplitudes;
• a feature extraction step of extracting sound quality featuring quantities, quantifying the sound quality, at an interval of said preset time length in the acoustic signals;
• a candidate domain evaluating step of evaluating, based on said short-term amplitudes and said sound quality featuring quantities, whether or not said candidate domain is said featuring portion;
• an index generating step of generating the index information including at least a start position and an end position of the candidate domain and the degree of crucialness of the candidate domain evaluated as being said crucial portion by said candidate domain evaluating means;
• and
• a recording step of recording said index information along with said acoustic signals.