JPH07170194A - Data coder - Google Patents

Abstract

PURPOSE:To reduce the adverse effect of an individual difference onto a human sensing characteristic by providing plural compressors eliminating a data component based on the human sensing characteristic and a selector selecting a desired compressor among the compressors to the code. CONSTITUTION:Digital music data are fed to a selector 1 that selects which compressor (1-n) is to be selected among plural compressors 2. As a result, the digital music data are fed to any selected compressor among the compressors 1-n. Each of the compressors 1-n in able to make compression by itself. Thus, in this coder, plural compression methods each is based on a different sensing characteristic are selected and the selector 1 selects one of the compression methods. The data are compressed according to the selected compression method, which generates compression data. Since the data component to be deleted is revised in the case of compression, the method copes with the change in the data component capable of being eliminated depending on each audible sense characteristic, difference from the preference and difference from the reproducing state or the like.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a data coding apparatus, and more particularly to data compression utilizing human sensory characteristics such as visual characteristics and auditory characteristics.

[0002]

2. Description of the Related Art In the field of digital audio, there were CDs, DATs and the like. These devices record physical data as they are. However, in reality humans are not aware of all this physical data. Also, the human ear does not have uniform sensitivity to sounds of all frequencies, some frequencies have good sensitivity and some have poor sensitivity, and the level of inaudible sound at each frequency is Called the minimum audible limit. This is the case with a standard human auditory characteristic, in which the sensitivity is the highest in the vicinity of 4 kHz, and a curve as shown in FIG. 14 is obtained. Also, when a sound of an arbitrary frequency is input to the ear, this effect of making a sound of a nearby frequency including the sound's own frequency inaudible is called masking. This looks like the curve in FIG. This curve is called a masking curve, and you cannot hear sounds smaller than this curve. Due to these two effects, a sound smaller than the curve as shown in FIG. 16 cannot be heard, so that the data component of this portion cannot be recognized by humans even if it exists or not. This curve is called a masking curve that takes the minimum audibility into consideration. By deleting the data component by utilizing the property that even if the unrecognized data component is deleted, there is no effect. Therefore, the data can be compressed at a high compression rate. As an apparatus for performing such data compression, MPEG (Moving Pictures)
ExpertGroup) / AUDIO, MD (Mi
ni Disk), DCC (Digital Comp)
act Cassette) etc.

[0003]

However, in the conventional method, compression is performed based on a single sensory characteristic model.
The variation of human sense is very large. For example, considering the minimum hearing limit, the minimum hearing limit of all humans is shown in FIG.
It is not represented by the curve of 4. Some humans are sensitive to low frequencies as shown in FIG. 17, some humans are sensitive to high frequencies, and masking has characteristics different from the masking curve modeled in FIG. Since some humans have different masking curves that take into account different minimum audible limits, as shown in FIG. Therefore, a human having the auditory characteristics described in this example senses the difference between the original sound and the compressed and expanded sound in the portion where the masking value considering the minimum audibility in FIG. 19 is smaller than the model. . Then, when trying to realize a high compression ratio, since the compression is performed around the boundary where a person who has a modeled sensory characteristic can recognize the difference or not, it becomes easier to detect the difference. There is also a psychological side to this perceptibility. In other words, even if the sensory organ itself has a perceptible difference, a situation may occur in which humans do not notice the difference. this is,
When a person listening to music, etc., concentrates his or her mind in the low and mid frequencies, it is a phenomenon that people can not hear even in the high frequencies, even though strange noises such as noise should be noticed.
Also, if you actually try to compress the music data while keeping the compression rate per unit time constant or by setting the lower limit of the compression rate, you can see that not only the data component that can be deleted but also the deletion is deleted. It may be necessary to delete such data components as well. In such a case, it is better to focus on individual tastes such as musicality, growth in the high range, and the lack of roughness in the middle and low ranges.
It will be easy for the listener to hear. This is because when listening to ordinary music, it is rare to listen to it while comparing it with the original sound, so it is better to hear the reproduced sound. By the way, since individual tastes vary widely, compression cannot be performed based on one sensory characteristic.

In such an algorithm, the data components that can be deleted differ depending on the reproduction status. For example, if there is stationary noise in the reproduced sound field, the component of the sound masked by the noise cannot be heard by the human ear,
Such data components can be deleted. Further, depending on the reproduced sound field, the signal of each frequency component changes due to the influence of echo, resonance and the like. In this way, when the magnitude of the signal component during reproduction changes, the masked data component also changes. If the signal component and the masking value considering the minimum audible limit are different, it is difficult to say that the signal component is already based on the sensory characteristic, and the data component necessary as the reproduced sound is deleted. Further, the change of the magnitude of the signal component at the time of reproduction is changed by increasing the low frequency signal or decreasing the mid frequency signal according to the taste of the listener.

[0005]

According to the data coding apparatus of the present invention as set forth in claim 1, data coding for compressing and decompressing digital data by deleting unrecognizable data components based on human sensory characteristics. In the device,
A plurality of compression devices having a function of deleting the data component based on the sensory characteristics, and a selection device for selecting a desired compression device from the plurality of compression devices are provided.

According to a second aspect of the present invention, there is provided a data encoding apparatus which performs compression / expansion by deleting unrecognizable data components of digital data based on human sense characteristics. A compression device having a plurality of means for performing data processing based on characteristics, and a selection device for selecting a desired means from the plurality of means are provided.

According to a third aspect of the present invention, there is provided the data encoding device according to the first or second aspect, wherein the data encoding device extracts the sensory characteristics of the user, and the extracting means. And a selection device that selects a desired compression device from a plurality of compression devices based on the extracted result.

A data coding apparatus according to a fourth aspect of the present invention is the data coding apparatus according to the first or second aspect, wherein an analyzing apparatus for analyzing a reproduction situation and an analysis from the analyzing apparatus are provided. And a selection device for selecting a desired compression device from the plurality of compression devices based on the result.

[0009]

The coding apparatus of the present invention is capable of switching a plurality of compression methods based on different sensory characteristics, and has a selection apparatus for selecting one of the compression methods. . The input digital signal is compressed according to the compression method selected by the selection device to generate compressed data. As a result, it becomes possible to change the data components that can be deleted during compression, and it is possible to deal with changes in the data components that can be deleted due to differences in individual auditory characteristics, tastes, differences in playback status, etc. Become.

[0010]

Embodiment 1 An embodiment in which a digital data encoding apparatus based on human sense characteristics according to the present invention is applied to music compression will be described in detail with reference to the drawings. FIG. 1 shows the configuration of the encoding device. FIG. 2 shows the configuration of the compression device shown in FIG. FIG. 3 shows the configuration of the decompression device for the compression device.

In FIG. 1, the digital music data is
It is sent to the selection device 1. The selection device 1 selects which of the compression devices 1 to n to use from the plurality of compression devices 2 and selects the compression device 1 to n for which digital music data has been selected.
Transfer to either. Each compression device is a device as shown in FIG. Next, the operation of this compression device will be described with reference to FIG. The digital music data is generated by the subband analysis unit 3 into frequency subband data. Scale factor calculator 4
Then, the scale factor of each subband is calculated from the size of the data of each subband. The masking calculation unit 5 calculates a masking value in consideration of the minimum audibility limit based on the sub-band data and the human auditory characteristics modeled by the compression device. The required number-of-bits calculation unit 6 calculates the number of bits required for each subband from the subband data and the masking value so that the modeled human cannot distinguish the compressed and expanded sound from the original sound. . This number of bits is called the required number of bits. The bit allocation unit 7 determines the number of bits actually allocated to each subband so as to suit the auditory characteristics of a person who has made a model, from the required number of bits of each subband, the total number of bits on the system, and the like. The quantizer 8 quantizes each subband data from the subband data, the scale factor, and the number of allocated bits to create compressed data. Then, the compressed data is decompressed by using the decompression device as shown in FIG. In FIG. 3, the compressed data is returned to each subband data by the dequantizer 9. The subband data is synthesized by the subband synthesis unit 10 to generate a reproduced sound. The details of the sub-band compression device here are described in JP-A-4-2.
50723.

As described in the prior art, the human auditory characteristic is represented by the minimum audibility limit and masking, and the compression device has data such as the minimum audibility limit and masking of the model, and the compression is performed based on the data. ing. In this encoding device, the plurality of compression devices are based on different auditory characteristics. Here, for the sake of explanation, a person named Mr. A will be taken as an example. First, the minimum audible limit will be described. A
It is assumed that his minimum audible curve has the frequency characteristic of FIG. At this time, it is assumed that the encoding device has compression devices having different characteristics of the minimum audible limit as shown in FIGS. Here, for simplicity of explanation, only four types will be described. This (a)
From among the curves in the minimum audible range from (d) to (c), a curve close to that of Mr. A is selected. Similarly, an appropriate masking curve is selected. Then, as shown in FIG. 6, compression is performed using a compression device based on a masking curve that considers the minimum audibility limit that is close to the masking curve that considers Mr. A's minimum audibility limit. In other words, for the same input data, since the respective compressors have different audible characteristics as models, the compressed data created by the respective compressors also differ. The data obtained by decompressing this compressed data using the decompressing device as shown in FIG. 3 differs depending on the compression device used. Here, if a person who listens to the decompressed data performs compression using a compression device based on an auditory characteristic model close to his or her own auditory characteristic, the deleted data component and the data component inaudible to the listener match. , It is possible to obtain a compressed / decompressed sound that is very difficult for the listener to distinguish from the original data.

Although the compression of music has been described here,
Even if music is replaced with an image, an encoding device having the same effect can be created.

Second Embodiment In the first embodiment, a plurality of compression devices are used to express some sensory characteristics and the like. However, there is a common part between these compression devices, which is common. It is possible to construct a compact encoding device. An embodiment of this device will be described in detail with reference to the drawings. Figure 7
The configuration of the encoding apparatus according to this embodiment is shown in FIG.

In FIG. 7, the digital music data is input to the subband analysis unit 11 of the encoding device. The subband analysis unit 11 generates subband data from digital music data. The scale factor calculation unit 12 calculates the scale factor from the size of the data of each subband. The selection device 13 selects which of the plurality of masking calculators 14 to use and transfers the subband data to the selected masking calculator. The selected masking calculator calculates a masking value from the auditory characteristics modeled by the subband data encoding device. Here, three masking calculation units are drawn to simplify the figure, but there are as many masking calculation units as necessary. The required bit length calculation unit 15 calculates from the subband data and the masking value how many bits each subband data has for human hearing. This number of bits is called the required number of bits. The bit allocation unit 16 determines the number of bits actually allocated to each subband from the required number of bits and the total number of bits limitation on the system. In the quantizer 17, the subband data,
Quantize each sub-band data from the scale factor and the number of allocated bits to create compressed data. By the way, although there are a plurality of masking calculation units 14, for the same input data, each masking calculation unit calculates the masking value in consideration of the minimum audible limit based on different auditory characteristics, so that it is actually selected. The data component is deleted based on the auditory characteristics modeled by the masking calculation unit. This makes it possible to realize a masking curve that considers a plurality of minimum audible limits for the same input, as in the case of using a plurality of compression devices in the first embodiment. If the compressed data is compressed using the masking calculator that models the decompressed data as a model of the auditory characteristics similar to the auditory characteristics of the listener, the deleted data component and the data component inaudible to the listener will match well. Therefore, it is possible to obtain a compressed / decompressed sound that is very difficult for a listener to distinguish from the original data.

Although only the masking calculation unit is selected here, other parts may be selected, and a method of correcting the value of the variable used in the calculation in each device is used. Things are possible. Further, although the description about the compression of music is made, an encoding device having the same effect can be made even if music is replaced with an image or the like.

Third Embodiment In the first and second embodiments, the selection of the selection device is performed manually, but in this selection, the model of the sensory characteristic which is the basis of the compression determines the ambiguity of the human sense. Difficult to include. Therefore, a device capable of selecting effective compression can be configured by adding a device for extracting the human sensory characteristic and automatically selecting it. FIG. 8 shows the configuration of the encoding device in this embodiment including the addition device. Figure 9
Shows a sensory characteristic extraction device. In addition, similar to the first and second embodiments, music will be described.

The individual human 18 has information on the characteristics of the sensory organs and the sensory characteristics which are the basis of compression such as psychological information. The information of the individual sensory characteristics is used as the sensory characteristic extraction device 1
Extract at 9. FIG. 9 shows this sensory characteristic extraction device.
In this apparatus, various sine wave test data are created by the sine wave sound source generator 22 or selected from the stored data. The target person 24 is made to hear the test sound using the acoustic device 23. The human 24 inputs into the input device 25 how the test sound sounds. Based on this data, the minimum audibility limit data creation unit 26 creates masking data in consideration of the minimum audibility limit. Here, if the test sound is a sine wave having a frequency, and by checking how loud the sound can be heard at each frequency, the value of the minimum audible limit can be checked. In addition, the approximate masking value can be examined by using a sine wave with two different frequencies as a test sound. Then, based on the extracted information such as the minimum audibility limit and masking, the optimum compression device selection device 20 selects the optimum one of the compression methods that the encoder with switching device 21 has. As an example of this selection method, the selection is made by selecting the one that has the smallest audibility limit of the sensory characteristics modeled by the listener and each compression device and the mitigation of the error at each frequency of masking. The encoder with a switching device is the encoding device shown in the first and second embodiments and the like, compresses the input digital music data based on the compression selected by the optimum compression device selection device, and outputs the compressed data. To do.

Although the compression of music has been described here,
Even if music is replaced with an image or the like, it is possible to make an encoding device having the same effect.

Embodiment 4 An embodiment in which a device in which a device taking into consideration a reproduction situation is added to a device for encoding digital data based on human sense characteristics according to the present invention is applied to music compression will be described in detail with reference to the drawings. explain. FIG. 10 shows the configuration of the encoding apparatus according to this embodiment.

In FIG. 10, the test sound generation device 27 generates test sounds such as sine waves of various frequencies for checking the reproduction status. This test sound is converted into spatial sound waves using the acoustic device 28. This spatial sound wave is transmitted to the sound field 29. This sound field 29 represents the reproduction situation of a human being who listens, and when a speaker system is used in a room, it is necessary to consider factors such as the spatial expansion and shape of the room, the reverberation by the wall material, and noise from the outside. If headphones are used, in addition to factors such as the shape of the headphones, the shapes of ears and face, noise from the outside, etc., it is possible to use them in the presence of others, so it is necessary to pay attention to sound leakage. Also, the sound to be heard is affected not only by the situation of this sound field but also by the characteristics of the acoustic device itself. Therefore, the sound field analysis unit 30 converts the reproduction characteristics into data based on the physical characteristics of the acoustic device and the analysis results using the sound ray method, the mirror image method, the finite element method, etc., which are conventionally used in the design of a hall. Also, the sound is actually picked up by the microphone 31 as to how the test sound sounds. In the optimum compression device selection unit 32, the difference between the sound picked up by the microphone 31 and the original test sound, the sound that becomes physically hard to hear or becomes inaudible from the data of the sound field analysis unit 30, the sound that is emphasized conversely, etc. Examine
The most suitable one of the compression devices of the encoding device with switching device 33 is selected. As an example of this selection method, if there is a data component that is not physically audible in the playback situation, the value of the minimum audibility limit is set to the value of the minimum audibility limit that is selected without considering the playback situation and the inaudible data component. There is a method of selecting a compression device having a minimum audible limit data close to the minimum audible limit data created by selecting the larger value. The encoding device with switching device 33 is the encoding device shown in the first and second embodiments, compresses the input digital music data based on the compression selected by the optimum compression device selecting device, and outputs the compressed data. To do.

If stationary noise as shown in FIG. 11 is generated in the reproduced sound field, a portion masked by the stationary noise occurs. Therefore, a sound smaller than the masking curve in FIG. 11 cannot be heard. this is,
Since it can be considered separately from the one that cannot be heard due to the auditory characteristics, in the part where the masking curve by the stationary noise is larger than the masking curve considering the minimum audible limit as shown in FIG. 12, the data which may be further deleted. The ingredients arise. Further, when the magnitude of the signal of each frequency is artificially changed by resonance in a reproduced sound field or an acoustic device, the magnitude of the signal component of each frequency changes. By the way, since the masking curve used for compression changes according to the signal, the curve changes as shown in FIG. If the data is compressed and expanded without taking this into consideration, the masking curve at the time of compression and the masking curve at the time of reproduction are deviated from each other, so that the data component necessary at the time of reproduction is deleted at the time of compression. At this time, if a compression device having a characteristic close to the changed masking curve is selected and compression is performed, the data component necessary for reproduction will not be deleted.

Although the compression of music has been described here,
Even if music is replaced with an image or the like, it is possible to make an encoding device having the same effect.

[0024]

According to the present invention, by selecting a compression device, it is possible to absorb individual differences in human senses without lowering the compression rate. As a result, in the current compression method using the human sensory characteristics, it is possible to reduce the number of humans who are uncomfortable by comparing the original data and the compressed / decompressed data.

[Brief description of drawings]

FIG. 1 is a diagram illustrating a configuration of an encoding device according to a first embodiment.

FIG. 2 is a diagram showing a configuration of a compression device according to the first embodiment.

FIG. 3 is a diagram showing a configuration of a decompression device according to the first embodiment.

FIG. 4 is a diagram showing a minimum audible limit characteristic of Mr. Human A according to the first embodiment.

FIG. 5 is a diagram showing an example of a model of a minimum audible limit of the compression apparatus according to the first embodiment.

FIG. 6 is a diagram showing a masking curve in which a minimum audibility limit of a model finally selected in Example 1 is taken into consideration and a masking curve in which a minimum audibility limit of Mr. Human A is taken into consideration.

FIG. 7 is a diagram illustrating a configuration of an encoding device according to a second embodiment.

FIG. 8 is a diagram illustrating a configuration of an encoding device according to a third embodiment.

FIG. 9 is a diagram illustrating a configuration of a sensory characteristic extraction device according to a third embodiment.

FIG. 10 is a diagram illustrating a configuration of an encoding device according to a fourth embodiment.

FIG. 11 is a diagram showing an example of stationary noise in a sound field and an example of a masking curve based on it.

FIG. 12 is a diagram showing an example of a masking curve in a noisy sound field and a masking curve in which a minimum audible limit is taken into consideration.

FIG. 13 is a diagram showing an example of a masking curve showing a change in the masking curve in consideration of the minimum audible limit when the signal changes.

FIG. 14 is a diagram showing a minimum audible limit.

FIG. 15 is a diagram showing a masking curve.

FIG. 16 is a diagram showing a masking curve in which the minimum audibility limit is taken into consideration.

FIG. 17 is a diagram showing an example of variation in characteristics of the minimum audible limit.

FIG. 18 is a diagram showing an example of a masking curve of a model and a masking curve of a listener.

FIG. 19 is a diagram showing an example of a masking curve in consideration of the minimum audible limit of a model and a masking curve of a listener.

[Explanation of symbols]

 1 Selection Device 2 Compressor 3 Subband Analysis Unit 4 Scale Factor Calculation Unit 5 Masking Calculation Unit 6 Required Bit Length Calculation Unit 7 Bit Allocation Unit 8 Quantization Unit 9 Inverse Quantization Unit 10 Subband Synthesis Unit

Claims (4)

[Claims] 1. A data encoding apparatus for compressing and decompressing digital data by deleting a data component which cannot be recognized based on human sensory characteristics, and a plurality of functions having a function of deleting data components based on sensory characteristics. And a selection device for selecting a desired compression device from the plurality of compression devices. 2. A data encoding device for compressing and decompressing digital data by deleting unrecognizable data components based on human sensory characteristics, and having a plurality of means for processing data based on sensory characteristics. A data encoding device comprising: a device; and a selection device for selecting a desired device from the plurality of devices. 3. A selection device for extracting a sensory characteristic of a user, and a selection device for selecting a desired compression device from a plurality of compression devices based on a result extracted by the extraction device. Claim 1 or Claim 2 characterized
The data encoding device according to. 4. An analysis device for analyzing a reproduction situation, and a selection device for selecting a desired compression device from a plurality of compression devices based on an analysis result from the analysis device. The data encoding device according to claim 1 or 2.