Signal processing method and apparatus, signal processing program, and recording medium

A signal processing method comprising:

• a playback step for playing back an audio signal recorded in recording means;
• a retrieval-position designating step for designating a desired retrieval position in the played-back audio signal;
• a spectral transformation step for transforming a portion of the audio signal in the retrieval position and retrieval part of the audio signal into time-frequency distributions;
• a dividing step for dividing the distribution corresponding to the portion in the desired retrieval position into a plurality of smaller regions;
• a parameter extracting step for extracting transformation parameters obtained when the smaller regions are transformed for establishing similarity to the retrieval part;
• a totaling step for totaling similarity values which are calculated based on the transformation parameters;
• a similarity evaluating step for, based on the totaled result, evaluating the similarity of the portion in the desired retrieval position to the retrieval part;
• and
• a similar-region detecting step for detecting each of the smaller regions of the portion in the desired retrieval position which is similar to the retrieval part.

A signal processing method comprising:

• a spectral transformation step for transforming a plurality of input audio signals into time-frequency distributions;
• a dividing step in which, among the time-frequency distributions corresponding to the audio signals, the distribution corresponding to at least one audio signal is divided into a plurality of smaller regions;
• a parameter extracting step for extracting transformation parameters obtained when transformation of the smaller regions is performed for establishing similarity to another audio signal;
• a totaling step for totaling similarity values which are calculated based on the transformation parameters;
• a similarity evaluating step for, based on the totaled result, evaluating similarities among the audio signals;
• a similar-section extracting step for detecting a similar section which is common to the distributions corresponding to the audio signals;
• a first encoding step for encoding the similar section detected in said similar-section extracting step;
• and
• a second encoding step for encoding a dissimilar section other than the similar section.

A signal processing apparatus comprising:

• spectral transformation means for transforming a plurality of input audio signals into time-frequency distributions;
• dividing means in which, among the time-frequency distributions corresponding to the audio signals, the distribution corresponding to at least one audio signal is divided into a plurality of smaller regions;
• parameter extracting means for extracting transformation parameters obtained when transformation of the smaller regions is performed for establishing similarity to another audio signal;
• totaling means for totaling similarity values which are calculated based on the transformation parameters;
• and
• similarity evaluating means for, based on the totaled result, evaluating similarities among the audio signals.

A signal processing apparatus comprising:

• recording means in which a plurality of audio signals are recorded;
• playback means for playing back the audio signals;
• retrieval-position designating means for designating a desired retrieval position in the played-back audio signals;
• spectral transformation means for transforming a portion of the audio signals in the desired position and retrieval part of the audio signals into time-frequency distributions;
• dividing means for dividing the distribution corresponding to the portion in the desired position into a plurality of smaller regions;
• parameter extracting means for extracting transformation parameters which are obtained when the smaller regions are transformed for establishing similarity to the retrieval part;
• voting means for totaling similarity values which are calculated based on the transformation parameters;
• similarity evaluating means for, based on the totaled result, evaluating the similarity of the portion in the desired position to the retrieval part;
• and
• similar-region detecting means for detecting the similarity to the retrieval part of each of the smaller regions corresponding to the portion in the desired position.

A signal processing apparatus comprising:

• spectral transformation means for transforming a plurality of input audio signals into time-frequency distributions;
• dividing means in which, among the time-frequency distributions corresponding to the audio signals, the distribution corresponding to at least one audio signal is divided into a plurality of smaller regions;
• parameter extracting means for extracting transformation parameters obtained when transformation of the smaller regions is performed for establishing similarity to another audio signal;
• voting means for totaling similarity values which are calculated based on the transformation parameters;
• similarity evaluating means for, based on the totaled result, evaluating similarities among the audio signals;
• similar-section extracting means for detecting a similar section which is common to the distributions corresponding to the audio signals;
• first encoding means for encoding the similar section detected in said similar-section extracting means;
• and
• second encoding means for encoding a dissimilar section other than the similar section.

A signal processing program comprising:

• a spectral transformation step for transforming a plurality of input audio signals into time-frequency distributions;
• a dividing step in which, among the time-frequency distributions corresponding to the audio signals, the distribution corresponding to at least one audio signal is divided into a plurality of smaller regions;
• a parameter extracting step for extracting transformation parameters obtained when transformation of the smaller regions is performed for establishing similarity to another audio signal;
• a totaling step for totaling similarity values which are calculated based on the transformation parameters;
• and
• a similarity evaluating step for, based on the totaled result, evaluating similarities among the audio signals.

A signal processing program comprising:

• a playback step for playing back an audio signal recorded in recording means;
• a retrieval-position designating step for designating a desired retrieval position in the played-back audio signal;
• a spectral transformation step for transforming a portion of the audio signal in the retrieval position and retrieval part of the audio signal into time-frequency distributions;
• a dividing step for dividing the distribution corresponding to the portion in the desired retrieval position into a plurality of smaller regions;
• a parameter extracting step for extracting transformation parameters obtained when the smaller regions are transformed for establishing similarity to the retrieval part;
• a totaling step for totaling similarity values which are calculated based on the transformation parameters;
• a similarity evaluating step for, based on the totaled result, evaluating the similarity of the portion in the desired retrieval position to the retrieval part;
• and
• a similar-region detecting step for detecting each of the smaller regions of the portion in the desired retrieval position which is similar to the retrieval part.

A signal processing program comprising:

• a spectral transformation step for transforming a plurality of input audio signals into time-frequency distributions;
• a dividing step in which, among the time-frequency distributions corresponding to the audio signals, the distribution corresponding to at least one audio signal is divided into a plurality of smaller regions;
• a parameter extracting step for extracting transformation parameters obtained when transformation of the smaller regions is performed for establishing similarity to another audio signal;
• a totaling step for totaling similarity values which are calculated based on the transformation parameters;
• a similarity evaluating step for, based on the totaled result, evaluating similarities among the audio signals;
• a similar-section extracting step for detecting a similar section which is common to the distributions corresponding to the audio signals;
• a first encoding step for encoding the similar section detected in said similar-section extracting step;
• and
• a second encoding step for encoding a dissimilar section other than the similar section.

A recording medium controllable by a computer, containing a signal processing program comprising:

• a spectral transformation step for transforming a plurality of input audio signals into time-frequency distributions;
• a dividing step in which, among the time-frequency distributions corresponding to the audio signals, the distribution corresponding to at least one audio signal is divided into a plurality of smaller regions;
• a parameter extracting step for extracting transformation parameters obtained when transformation of the smaller regions is performed for establishing similarity to another audio signal;
• a totaling step for totaling similarity values which are calculated based on the transformation parameters;
• and
• a similarity evaluating step for, based on the totaled result, evaluating similarities among the audio signals.

A recording medium controllable by a computer, containing a signal processing program comprising:

• a playback step for playing back an audio signal recorded in recording means;
• a retrieval-position designating step for designating a desired retrieval position in the played-back audio signal;
• a spectral transformation step for transforming a portion of the audio signal in the retrieval position and retrieval part of the audio signal into time-frequency distributions;
• a dividing step for dividing the distribution corresponding to the portion in the desired retrieval position into a plurality of smaller regions;
• a parameter extracting step for extracting transformation parameters obtained when the smaller regions are transformed for establishing similarity to the retrieval part;
• a totaling step for totaling similarity values which are calculated based on the transformation parameters;
• a similarity evaluating step for, based on the totaled result, evaluating the similarity of the portion in the desired retrieval position to the retrieval part;
• and
• a similar-region detecting step for detecting each of the smaller regions of the portion in the desired retrieval position which is similar to the retrieval part.

A recording medium controllable by a computer, containing a signal processing program comprising:

• a spectral transformation step for transforming a plurality of input audio signals into time-frequency distributions;
• a dividing step in which, among the time-frequency distributions corresponding to the audio signals, the distribution corresponding to at least one audio signal is divided into a plurality of smaller regions;
• a parameter extracting step for extracting transformation parameters obtained when transformation of the smaller regions is performed for establishing similarity to another audio signal;
• a totaling step for totaling similarity values which are calculated based on the transformation parameters;
• a similarity evaluating step for, based on the totaled result, evaluating similarities among the audio signals;
• a similar-section extracting step for detecting a similar section which is common to the distributions corresponding to the audio signals;
• a first encoding step for encoding the similar section detected in said similar-section extracting step;
• and
• a second encoding step for encoding a dissimilar section other than the similar section.