JPH0231393A - Method and device for data editing - Google Patents

Abstract

PURPOSE:To easily and quickly retrieve specific voice data by using features extracted from voice data as a part of management data and pairing management data with plural voice data and storing it in a storage part. CONSTITUTION:A feature extracting stage 2 where voice data is processed to obtain a part of management data and a management data generating stage 3 where a part of management data is used to generate management data pairing with voice data are provided, and management data is stored. The feature extracting part 2 detects an average level of voice data and segments parts having a certain value or larger for an arbitrary section and compresses the time base thereafter to generate a part of management data. Thus, since the beginning of plural voice data is continuously reproduced for n-number of seconds at the time of obtaining a list of plural storage voice files, a required voice file is easily and quickly specified.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of Industrial Application The present invention relates to a data editing device that inputs a large amount of data such as audio data into a computer as an audio data file and edits a plurality of audio data files.

BACKGROUND OF THE INVENTION In recent years, as the number of devices that can handle audio signals as digital data has increased, there has been an increasing desire to perform editing processing on a computer. In order to edit data quickly and easily, it is most practical to input audio data into a computer and edit it by making full use of functions such as arithmetic processing. method is used.

Hereinafter, a conventional data editing device as described above will be explained with reference to the drawings.

FIG. 2 shows the configuration of a conventional data editing processing device.

In FIG. 2, 21 is an input section for inputting audio data from the outside, 23 is a storage section for storing audio data, 22 is a management section for managing the audio data stored in the storage section 23, and 24
25 is an output unit that outputs the audio data read by the management unit 22 from the storage unit 23 to the outside, and 25 is the input unit 21 and the management unit 22.
and the output unit 24, and 26 is a control command input unit that inputs control commands to the control unit.

The operation of the data editing device configured as described above will be explained below.

First, according to an input command input from the control command input section 26, audio data from the outside is inputted to the input section 21 and then sent to the management section 22. The management section 22 converts the audio data sent from the input section 21 into a file and stores it in the storage section 23.

Next, according to the output command input from the control command input section 26, the management section 22 reads out the audio data file from the storage section 23 and outputs it to the output section 24, and the output section 24 outputs the audio data to an external device such as a speaker. do.

Problems to be Solved by the Invention However, with the above configuration, when a large number of audio data files exist in the storage unit 23, the only information for searching is the file name, and it is difficult to quickly search for the desired audio file. was extremely difficult.

In view of the above-mentioned problems, the present invention provides a data editing method and a data editing device that can easily find specific audio data from a plurality of audio data stored in a storage unit.

Means for Solving the Problems In order to solve the above problems, the data editing method of the present invention includes an input process of inputting audio data from the outside, and processing of the audio data obtained in the input process to create part of the management data. a management data creation process of creating management data paired with audio data using a part of the management data; and a storage process of storing the audio data and management data;
It consists of a feature restoration process that restores management data to part of the audio data, and an output process that outputs part of the restored audio data to the outside.Furthermore, the feature extraction process detects the average level of the audio data and sets it to a constant value. The feature is that a portion of the management data is created by cutting out an arbitrary section from the above portion and then compressing the time axis.

In addition, a data editing device that embodies this method includes a storage unit that stores a plurality of audio data and a plurality of management data paired with these, and a storage unit that creates management data and inputs and outputs the audio data and management data to the storage unit. At the same time, there is a management section that manages audio data by referring to management data, an input section that is connected to this management section and inputs audio data from the outside, and an input section that is connected to the management section and processes the audio data obtained from the input section. A feature extraction unit that generates part of the management data, an output unit that is connected to the management unit and outputs the audio data retrieved from the storage unit, and an output unit that outputs the audio data retrieved from the storage unit that is connected to the management unit. a feature restoration unit that processes a part and outputs it to an output unit;
It has a configuration including a management section, an input section, a feature extraction section, an output section, a control section that controls the feature restoration section, and a control input section that inputs control commands to this control section from the outside. It is characterized by detecting the average level of the audio data, cutting out an arbitrary section of the part where the level exceeds a certain value, and then compressing the time axis to create a part of the management data.

Effect of the Invention With the above-described configuration, the present invention extracts features from input voice data and makes them part of management data, and stores a plurality of voice data and management data that are managed in pairs with these data in a storage unit. Therefore, when searching for specific audio data, it is possible to easily and quickly find specific audio data by restoring the characteristics of the management data and outputting a portion of the audio data to an external device for confirmation.

Embodiment Hereinafter, a data editing apparatus according to an embodiment using the data editing method of the present invention will be described with reference to the drawings.

FIG. 1 shows a configuration diagram of a data editing device in an embodiment of the present invention.

In Fig. 1, 1 is an input unit that inputs audio data from the outside, 2 is a feature extraction unit that extracts features from the input audio data and makes them part of management data, and 4 stores audio data files and management data files. A storage unit 3 stores the audio data obtained from the input unit and part of the management data obtained from the feature extraction unit into an audio data file and a management data file.
6 is an output unit for receiving and outputting the data which the management unit 3 retrieves the audio data file from the storage unit 4 and outputs as audio data to the outside; 6 is the management unit that retrieves the audio data file from the storage unit 4; A feature restoring unit that restores the features of audio data from a part of the data; 7 is an input unit 1 a feature extraction unit 2 a management unit 3 a feature restoring unit 6 a control unit that controls the output unit 6;
is a control command input section that inputs commands to the control section from the outside.

The operation of the data editing device configured as described above will be explained below.

First, when a command "1 voice data input" is input from the control command input section 8 to the control section 7, the control section 7 issues control to the input section 1, feature extraction section, and management section 3.

The input section 1 inputs audio data from the outside and sends it to the management section 3. The management section 3 stores this audio data in the storage section 4 as a file. The feature extraction unit 2 extracts the input audio data (
Detect the level in Figure 3 a) Figure 3 b) Cut out the data for n seconds from the point where it exceeds a certain level /L/L Figure 3 C) Resample this data into 1 part m d) and sends it to the management section 3 in a compressed time-axis manner. Reply here/
l/L, interval unit n, and resampling variable 1/m are controlled by the control unit 7.
You can set it as you like. The management section 3 receives a part of the management data from the feature extraction section 2, combines it with information given from the control section t, and stores it in the storage section 4 as a management data file. An example of the management data file is shown in FIG. The management unit 3 always manages the audio data file and the management data file as a pair of files. As a method of managing the files as a pair, for example, a part of the file name is made common. A method such as writing the storage location of the audio data file μ in the management data file is used. By repeating this operation, a plurality of audio data files and management data file μ are stored in the storage section 4.

Next, when the command for 1 voice data list 1 is inputted to the control section 7 from the control command input section, the control $7 issues control to the management section 3, feature restoration section 6, and output section 6. The management section 3 extracts the management data file from the storage section 4 and sends it to the feature restoration section 6. The feature restoring unit 6 restores the feature data extracted and recorded from the audio data at the time of input from the management data into audio data and sends it to the output unit 6. The output unit 6 outputs the data from the feature restoration unit 6 to an external device such as a speaker. The audio data restored by the feature restoration unit 6 is the first n seconds of audio data stored in the storage unit 4, and is compressed in the time axis. The feature restoring unit 6 controls if necessary.
7, the time axis is expanded or the pitch is converted using an appropriate interpolation method such as linear interpolation. Since the first n seconds of the audio data are continuously reproduced externally, the required audio file can be easily identified and recognized.

In this embodiment, compression was performed by resampling when extracting features from audio data, but compression may also be performed using a speech synthesis analysis method, such as PARCOR.

In addition, the management data file used in this example includes:
In addition to the feature data extracted from the voice data, search keywords may be input from the control command input section 8 and used.

In this embodiment, since the audio data recorded and reproduced as management data is compressed and expanded, the quality is degraded compared to the original audio data, but there is no problem when used for search purposes.

Effects of the Invention As described above, according to the present invention, there is provided a storage section that stores a plurality of audio data and a plurality of management data paired with these, and a storage section that creates management data and inputs the audio data and management data into the storage section. A management section that outputs and manages audio data by referring to management data, an input section that is connected to this management section and inputs audio data from the outside, and an input section that is connected to the management section and processes the audio data obtained from the input section. a feature extraction section that generates part of the management data; an output section that is connected to the management section and outputs the voice data retrieved from the storage section; and a feature extraction section that is connected to the management section and outputs the audio data retrieved from the storage section. A feature restoration section that processes a part of the data and outputs it to the output section, a control section that controls the management section, input section, feature extraction section, output section, and feature restoration section, and inputs control commands from the outside to this control section. The feature extraction section detects the average level of the audio data, cuts out a section of the audio data that exceeds a certain value, and then compresses the time axis to extract part of the system management data. By creating this, when you request a list of multiple stored audio files, the first n seconds of multiple audio data will be played back in sequence, allowing you to easily and quickly identify the required audio file. .

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the structure of a data editing device according to an embodiment of the present invention, FIG. 2 is a block diagram showing the structure of a conventional data editing device, FIG. 4
The figure is a data diagram showing an example of a management data file. 1°°゛°°° input section, 2...feature extraction section, 3
...Management section, 4.. Storage section, 6. Extraction section, 6. Characteristic extraction section, 7.. Control section, 8.. ...Control command input section. Name of agent: Patent attorney Shigetaka Awano and one other person

Claims (4)

[Claims] (1) An input process of inputting voice data from the outside, a feature extraction process of processing the voice data obtained in the input process to obtain a part of management data, and a process of extracting features from the voice data using a part of the management data. a management data creation process for creating a pair of management data; a storage process for storing the audio data and the management data; a feature restoration process for restoring the management data into a part of the audio data; and a restored audio A data editing method that consists of an output process that outputs part of the data externally. (2) The feature extraction step is characterized in that the average level of the audio data is detected, a part of the voice data that exceeds a certain value is cut out as an arbitrary section, and then a part of the management data is created by compressing the time axis. Data editing method described. (3) a storage unit that stores a plurality of audio data and a plurality of management data paired with the audio data; a storage unit that creates the management data, inputs and outputs the audio data and management data to the storage unit, and references the management data; a management section for managing the audio data; an input section connected to the management section for inputting audio data from the outside; and an input section connected to the management section for processing the audio data obtained from the input section and inputting the management data. an output unit connected to the management unit and outputting the audio data retrieved from the storage unit to the outside; and a feature extraction unit connected to the management unit and management data retrieved from the storage unit. a feature restoration unit that processes a part of the data and outputs it to an output unit; a control unit that controls the management unit, the input unit, the feature extraction unit, the output unit, and the feature restoration unit; A data editing device consisting of a control input section for inputting control commands. (4) The feature extraction unit further includes a detection unit that detects the average level of the audio data, a cutting unit that cuts out a section where the output of the detection unit is above a certain value, and compresses the extracted audio data on the time axis. 4. The data editing device according to claim 3, further comprising a time compression section.