JPS6330031A - Audible tone signal transmission system - Google Patents

Abstract

PURPOSE:To prevent the deterioration in sound quality in the audible sense even with an increased compression rate by applying the level correction in response to an input amplitude level in the unit of blocks according to a charac teristic inversed to the Fletcher-Munson equi-loudness curves, coding the input and sending it, restoring the amplitude level by a characteristic inverse to the former characteristic so as to listen to the sound after decoding. CONSTITUTION:An input signal being an electric signal converted and amplified from an audible sound signal is inputted to a selection amplifier circuit 2, where the signal is amplified or compressed. Then an output of each selection amplifier circuit is stored tentatively in a buffer memory circuit 3. Peak hold circuits 4-1-4-4 hold the peak value of the output of each selection amplifier circuit. A maximum position detection circuit 5 reads a value held in the peak hold circuit 4 at each prescribed time, selects one of output data within a prescribed time of the selection amplifier circuit, and a signal coding circuit 7 applies quantization and coding. The coded data is sent to a signal decoding circuit 9 and a selection decoding circuit 10, where the data is decoded. The decoded input signal is inputted to an inverse selection amplifier circuit 11, one of the outputs is selected and given to an output terminal 12.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to an audible sound signal transmission system, and particularly to an audible sound signal transmission system that obtains a high information compression rate without deteriorating the perceptual sound quality.

[Conventional technology]

There is an increasing demand for high-quality digital transmission of audible sounds such as voice and music along with images such as digital image memories and digital video telephones. To meet this demand, digital transmission of audible sounds using pulse, code modulation,
ion: hereinafter referred to as PCM) or even more efficient adaptive differential pulse code.
Adaptive Differentiation
tial Pulse Code Modulation
n; hereinafter referred to as ADPCM), adaptive
Transform Coding (Adaptive
Transform Coding (hereinafter referred to as ATC) has been researched and put into practical use.

For example, a well-known example of high-quality PCM is the compact
Considering a compact disc (hereinafter referred to as CD), the amount of information is approximately 700Kbits/S.
(44,1KElz sampling, 16-bit linear P
CM encoding), which makes it difficult to apply to digital videophones and the like.

As a high quality ADPCM, Masahiro Taka (Mas
Proceedings by Mr. Ahiro Taka et al.
oceed-ingI CAS S P 86) Magazine 19
Published in 1986, Volume 2, pp. 817-820
ITT Standard Digi/G Activities on Speech Coding J (C
CITT Standardizing Activit
A basic explanation of ATC is provided in the paper titled ES on 5peech Coding, Transaction Coding, by Rayner Zelinski et al.
On acoustic speech and signal processing volume knee varnish varnish
Bee 25. Number 4. August 1977 (IEE
ETrans, Acoustics, 5peec
h and Signal Processing,
Vol.

AS 5P-25, Nn4, August 197
7), pp. 299-309, 'Adaptive-Transform Coding on Speech Signals'
Coding of 5peech Signals
) is stated in the paper entitled.

With these highly efficient methods, it is possible to compress information to around 315 without deteriorating the sound quality compared to the simple linear PCM encoding described above, but if the compression rate is further increased, quantization noise increases and the sound quality deteriorates. deteriorates.

In the linear PCM encoding used for conventional CDs, the necessary number of bits is determined solely from the dynamic range of physical sound pressure (amplitude). Now, I
KHz, 40dB signal (pure tone) A and 100Hz
, 62 dB signal (pure tone) B is mixed. Individual signals are perceived by humans at the same loudness. For example, if you connect the IKElz signal A to a PC
Assuming that 9 bits are required for high precision (high quality) encoding with M, the mixed signal C
To encode 22 (62-40=22) dB
Since the signal amplitude of 100flz is large, a total of 13 bits, including the previous 9 bits plus 4 (22÷4 out of 6) bits, are required.

As a method for eliminating such problems, a pre-emphasis method is known in voice analysis, recognition or recording/playback systems. This is because the spectrum of the audio signal has a falling characteristic of 6 to 10 dB10 ct from around 800 Fiz due to the radiation characteristics of the mouth, so in order to correct this, a fixed rising characteristic of 6 dB 10 ct, for example, is given, and the high frequency and low frequency This is to prevent deterioration of sound quality due to analysis accuracy due to a small number of bits or quantization error.

However, with only the rising characteristic, 10KE like a music signal
If there is a signal with a high frequency higher than Iz, the fifth
5KH as seen from the equal loudness curve shown in the figure.
Inconveniences occur because the auditory sensitivity is poor even for frequencies higher than z.

For example, the sound pressure of IKI (z, 40 dB) feels the same as the sound pressure of 10 KEIz, which has a sound pressure of 53 dB.The signal obtained by mixing these two signals is 6 dFl/o.
As shown in Figure 6, the current difference is 13 (53-40=13) dR when the increasing characteristic of ct is multiplied.
is further increased by 20 dB, resulting in a signal of 33 dl13. Therefore, as explained above, if 9 bits are required to encode an I KHz signal with high accuracy using PCM, then the mixed signal requires an additional 6 bits (36 ÷ 6 out of 6), or 15 bits. becomes.

[Problem that the invention seeks to solve]

As explained above, in the above conventional technology, when the compression ratio is increased, the sound quality deteriorates. This is because the computerization accuracy is determined simply from a physical measure, such as the dynamic range of amplitude, without considering the characteristics of human hearing.

An object of the present invention is to take human auditory characteristics into consideration during quantization and encoding, and to prevent perceptual sound quality from deteriorating even if the compression rate is increased.

[Means for solving problems]

The above purpose is to perform level correction on an audible sound signal by adapting it to the input amplitude level on a block-by-block basis using the inverse characteristics of Fletcher-Munson's equal loudness curve or characteristics equivalent to it, encode it, and transmit it. do. This is achieved by listening to the signal with the amplitude level restored to its original level with a characteristic that is opposite to the previous characteristic after the signal has been encoded and decoded.

Figure 5 shows Fletcher Munson's equal loudness curve. (IE Child Communication Society edition, “Hearing and Speech” 10th edition, 1st
(Page 04) The vertical axis is sound pressure, and the horizontal axis is frequency. The numbers in the figure are the sound pressure of IKHz pure tones, that is, the loudness level (hon value), and the numbers in parentheses are the loudness (hon value).

For example, K, which perceives the same loudness (sound size) as a sound with I KHz and 40 dB sound pressure, is 10011
62dB at z.

At 3 KHz, the sound has a sound pressure of 37 dBO, and all pure tones along this curve have the same loudness level of 40 phon.

This equal loudness curve, for example, a 40-phone curve, shows the following.

(1) Human hearing is highly sensitive to sounds of 1 to 5K11z. Specifically, I KHz, 40 dB sound pressure (
For electrical signals, signal amplitude @) and 1001'lz, 62d
Since the sound pressure of B sounds the same loudness, IKH
z has 22dB higher sensitivity than 100 Fiz.

(2) Considering the normal human hearing range of 10 to 90 phons, the physical dynamic range of sound pressure is the same for all frequencies, but the dynamic range of human hearing, that is, the loudness, is 100 Hz. Then the sound pressure is 40d
50 (9O-40=5
0) dB, IKHz is 80 (90-10=8
0) Different in dB and frequency. FIG. 7 shows the relationship between this dynamic range and frequency. The dynamic range below I KHz decreases in proportion to the frequency, and I KHz
Above, it is constant. This shows that if the quantization can accurately reproduce the dynamic range of loudness at frequencies above IKHzKHz, the above quantization is sufficient for frequencies below IKHz.

Next, consider how many bits should be used to quantize the dynamic range of loudness.

Generally, even if the physical intensity of the applied stimulus changes, sensory organs cannot perceive the change unless the difference in intensity due to the change exceeds a certain value. Applying this to the auditory sound intensity (loudness), even if the physical sound pressure that is the stimulus being applied changes, if the sound pressure difference due to the change does not exceed a certain value, the sound will not be heard. I cannot perceive that the strength has changed.

There is a certain functional relationship between the intensity difference between stimuli and the probability of detecting the difference, and the intensity difference for which the probability of detecting the difference is 1/2 is defined as the discrimination threshold (Difference Limen:
D.L.

). The intensity difference between the stimuli corresponding to D and L is just noticeable.
difference: Jnd).

This Jnd is expressed as a physical quantity called sound pressure, for example, but it can be considered as a projection of a psychoacoustic quantity called sound intensity onto a physical measure called sound pressure. Within the physical scale of sound pressure, it is possible to measure the scale finely, but from the auditory perspective of sound intensity, divisions finer than a certain resolution are meaningless.In other words, the sound strength scale, L, is psychological. In terms of domain resolution, this corresponds to quantizing the physical domain of sound pressure.

FIG. 8 shows the relationship between the sound intensity, L, (Δ■) and the ratio discrimination range (ΔI/I) (T is the sensory level) and frequency. The curve groups 1 to 7 in the figure each represent the sensory level (
This is the case when I) is 5 to 80 dB. Figure 9 is the 8th
The diagram has been redrawn to show the relationship between D, L, and sensory level (I).

In Figures 8 and 9, the sensory level (5ensatio
nlevel ) I is defined as the value of the minimum audible range 0 phon in FIG. 5 as OdB.

From Figure 8, the sensory level ■ is 30 dB or more, and the frequency is 12.
If it is 8 FIz or more, L, (ΔI) is constant, and as the sensory level becomes larger, L, (ΔF) becomes smaller and becomes larger at both ends of the frequency, and L-(ΔI) becomes larger.

From Figure 9, it can be seen that when the sensory repelue is 60 dB or more, the ratio discrimination range (
ΔI/I) takes a nearly constant value, and Weber's law "ΔI/I=R: In other words, D and L are proportional to the level of the physical quantity (sound pressure) of the stimulus" holds true. .

The following points can be derived from the above results.

(3) Similar to the conclusion in section (1) above, the auditory resolution, that is, the sound intensity, L, is small when the sensory level ■ is small (
30 dB or less), that is, the sound pressure is small, that is, the amplitude of the electrical signal is small, and becomes high at 1 to 5 KHz.

That is, when the amplitude is small, signals outside the 1-5 KHz band may be quantized more coarsely than signals within the 1-5 KHz band.

In other words, if 1 to 5KH7 are quantized with high precision, bands other than the above should be quantized with sufficient precision if the number of quantization bits is the same.

(4) If the sensory level is 60 dB or higher, the ratio discrimination threshold takes a nearly constant value. In other words, the resolution that humans can perceive remains constant regardless of their sensory level. In other words, the sound pressure above a certain level, that is, the amplitude of the electrical signal, may be quantized at a certain level regardless of its level.

For example, both the amplitude of 60 dB and the amplitude of 80 dB may be quantized and encoded using the same number of bits.

Next, if we know the sound pressure that can be perceived by the sense of hearing and its deviation, L, we can find the number of steps that can be taken to analyze the sound pressure.

For example, let us consider the accumulation of pure tones of IKFlz from the minimum audible value OdB to the maximum audible value 120 dB, L, and beat.

In reality, the lower the level, the larger L becomes, so it becomes even smaller.

In other words, even in a wide range of 0 to 120 dB, the number of steps that can be analyzed aurally is at most 280 steps. Linear PCM requires only about 8 pits of information.

Considering the normal listening range of 10 to 90 phones, the cumulative n is about 118, and even with linear PCM encoding, the amount of information is about 7 bits.

In terms of the dynamic range and resolution of hearing,
It can be seen that the number of quantization bits in CD is too large.

This is because, as explained above, the number of quantization bits is determined solely from the physical dynamic range of the signal amplitude.

The present invention is based on the above-mentioned auditory characteristics, and its purpose is to provide a system that maintains quality that does not change the auditory sense even when information is compressed (simply, the number of quantization bits is reduced). There is something to do.

For this reason, (1) By passing an audible sound signal through the inverse characteristic of the Flexure Munson's equal loudness curve, its physical amplitude is projected onto the amplitude as a psychological measure.

(2'; The projected amplitude is quantized and encoded with the number of steps of psychological resolution.

0) Decode the signal and return it to its original physical amplitude by passing it through the Flexure Munson equal loudness curve.

(2) Since the shape of the Flexure Munson equal loudness curve differs depending on the sound pressure level of the audible sound, the b1 audible signal series is divided into blocks 17 at regular time intervals, and an optimal equal loudness curve is selected for each division unit.

Perform processing 2.

Specifically, the following processing is performed.

(1) Frequency-dependent amplitude expansion or compression is performed on the input audible sound signal using signal processing circuit means having characteristics and gain inverse to the Flexure Munson's equal loudness curve or its approximate curve. For example, 100Hz,
62 dB amplitude and I KRZ, 40 dB amplitude and 10
Considering an audible sound signal in which three types of pure tones (sine waves) with amplitudes of 100 FIz, 1 KHz, and 10 KEIz are mixed, the amplitudes of pure tones of 100 FIz, 1 KHz, and 10 KEIz become 40 dB, respectively. In other words, the amplitudes of each frequency at the same loudness level are all compressed into the same amplitude.

(2) A plurality of discrete circuit means are provided, and one of the outputs of the circuit means is selected to have a maximum output signal without clipping.

(3) This signal is quantized and encoded (straight line P CM ) t using a finite number of bits, for example 7 bits, and then transmitted. At the same time, selection information is also encoded and sent as sub information to indicate which signal processing circuit means output is to be encoded.

(4) At the same time as decoding the encoded data of the transmitted signal, the selection information of the previously selected signal compression circuit means is also decoded.

(5) A plurality of inverse signal processing circuit means are paired with the signal processing circuit means and have opposite characteristics, and one of the circuit means is selected based on selection information.

(6) The decoded signal is decompressed by the selected inverse signal processing circuit means and listened to.

[Effect]

Adaptive level correction on a block-by-block basis based on the inverse characteristic of Fletcher Munson's equal loudness curve raises the audible sound in the range of 1 to 5 KHz, for which human hearing sensitivity is high, to a relative maximum, and improves the S/N ratio in this band. Work to improve.

As a result, the number of quantization bits required for PCM encoding of this audible sound is the minimum necessary to reproduce the signal amplitude with high precision, and unlike conventional methods, it is possible to reproduce large amplitude signals in bands with low auditory sensitivity. The number of quantization bits can be made smaller than that required for quantization, and the amount of information can be compressed. Furthermore, even with such compression, the perceptual sound quality does not deteriorate.

〔Example〕

An embodiment of the present invention will be described below with reference to FIG. 1st
In the figure, 1 is an input terminal, 2-1 to 2-4 are selective amplifier circuits having frequency gain characteristics shown in FIG. 10, and 3-1 to 3-
4 is a buffer memory circuit for storing signals at fixed time intervals; 4-1 to 4-4 are peak hold circuits for storing signal peak values; 5 is for reading out the peak values of each peak hold circuit at fixed time intervals and storing the signals; a maximum position detection circuit that detects whether the signal is clipped or not and also detects the largest one among the non-clipped signals; [5 is a selection circuit that selects one of the outputs of the backup memory circuit; 7 is a signal 8 is a selection signal encoding circuit that encodes the selection signal of the maximum position detection circuit;
9 is a signal decoding circuit for decoding encoded signals, 1
0 is a selection signal decoding circuit that decodes the encoded selection signal, and 11-1 to 11-4 are selection amplification circuits 2-1 to 2-.
12 is an output terminal of an inverse selection amplifier circuit having characteristics opposite to those of 4;

The curves shown in FIG. 10 are selected amplifier circuits 2-1 to 2- from the top.
This is the frequency gain characteristic of 4. These are: ・Of the Fletcher Munson curve shown in Figure 35, sensory level 20,
It has completely opposite characteristics to the curve for 40, 60, and 80 phons, and using the inverse characteristic of the curve for 80 phon as 2-4 as a reference (approximately OdB as gain), the inverse characteristic of the curve for 60 phon is 2-3.The inverse characteristic of the curve at 40 phons is 2-
The inverse characteristic of the curve at 2.20 phons is set as 2-1. Characteristics 2-1 to 2-4 each have a gain difference of 20 dB at 4KElz.

An audible sound signal collected by a microphone or the like is converted into an electrical signal, amplified, and input to the input terminal 1 .

Now, as this input signal, consider an electrical signal equivalent to the sound pressure of a sensory level of 0 to 90 phon. When a 4KHz, 90-phone pure tone signal is input to the reference selection amplifier circuit 2-4, the output amplitude is the maximum (hereinafter referred to as vrwt), and any signal amplitude greater than this will be affected by the power supply voltage of the selection amplifier circuit or AD conversion. Assume that the output is clipped by the reference voltage of the circuit.

The input signals are simultaneously input to the selective amplification circuits 2-1 to 2-4, and are each subjected to amplitude expansion (amplification) or compression according to the characteristics shown in FIG.

For example, when a 4KHz, 80-phone signal is input, the output of the selective amplifier circuit 2-4 is -10dB compared to wx.
Although it is at the lower amplitude level and is not clipped, the output of 2-3 has a gain of 20 dB, so it becomes VW +10 dB and is clipped. The output of 2-2.2-1 is also clipped in the same way.

For example, if a pure tone of 100[1z, 40 phons is inputted, (although this has the same sound pressure level or amplitude level as a pure tone of 4KHz, 60 phons), the selection amplifier circuit 2-4.2- The output of 3 is not clipped and the output of 2-
2.2-1 output is clipped.

The output of each selection amplifier circuit is the backup memory circuit 3-1 to 3-3.
-4 is temporarily stored. At the same time, the peak values of the outputs of the respective selective amplifier circuits are held in the peak hold circuits 4-1 to 4-4. For example, if the selection amplifier circuit performs analog processing, memory is achieved by sampling this output signal, performing AD conversion, and storing this data in RAM.

Of course, the selective amplifier circuit may be digitally processed, and in this case, processing such as sampling, l) conversion, etc. is required between the input terminal 1 and the selective amplifier circuit.

The maximum position detection circuit 5 includes each peak hold circuit 4-1 to 4-4.
The value held at -4 is read at fixed time intervals, and one of the output data of the selection amplifier circuit stored in the buffer memory circuit within the fixed time period is selected. This selection is performed by searching for the largest one among the hold values that are not clipped.

For example, in the case of the aforementioned 100 Hz, 40 phone sound, the buffer memory circuit 3-3 connected to the selective amplification circuit 2-3 is selected.

The data in the selected buffer memory circuit is input to the signal encoding circuit 7 via the selection circuit 6, where it is quantized and encoded with a predetermined number of bits. For example, it is linearly PCM encoded with 7 bits.

At the same time, the selection signal from the maximum position detection circuit 5 is also encoded by the selection signal encoding circuit 8.

As mentioned above, the amplitude input to the signal encoding circuit 7 is
Regardless of the signal amplitude applied to the input terminal l, the amplitude is expanded or compressed so that the peak value is approximately the 7w value. Whether it is expansion processing or compression processing depends on the frequency and amplitude level of the input signal. When the input signal is small, it is amplified, that is, the amplitude is expanded.
Due to the characteristics shown in Figure 0, the 1 to 5 kHz band with high auditory sensitivity is relatively elevated compared to other bands.

Furthermore, the input signal is divided into blocks at regular time intervals, and the above expansion or compression processing is performed for each block, and one selection signal is attached to each block.

The encoded data is transmitted to a signal decoding circuit 99 and a selection signal decoding circuit 10 via respective transmission paths. Although the transmission paths are separated in this embodiment, the signals may be multiplexed and transmitted over a single transmission path. Furthermore, the method is not limited to the transmission path, and may be a method in which encoded data is first recorded on a medium such as a source disk, and later reproduced and inputted to a decoding circuit.

The encoded data is decoded by a signal decoding circuit 91 and a selection signal decoding circuit 10. The decoded input signals are simultaneously input to inverse selection amplifier circuits 11-1 to 11-4.

The inverse selection amplifier circuits 11-1 to 11-4 correspond to the selection amplifier circuits 2-1 to 2-4, respectively, and have characteristics (+60 dB) that are the opposite of the characteristic curve shown in FIG.
is -60 dB).

With this customization, previously expanded amplitudes are compressed, and conversely compressed amplitudes are expanded. By this operation, the signal whose amplitude has been compressed or expanded by the 8-selection amplifier circuit is returned to the original input signal.

One of the outputs of the inverse selection amplifier circuit is selected by the selection circuit 6 using the decoded selection signal and outputted to the output terminal 12. For example, the encoded selection signal is
3 &, the selection at this time is the output of the inverse selection amplifier circuit 11-3.

In the above explanation, as shown in Figure 10, the complex characteristics of the selective amplifier circuit that trace the Fletcher-Manson curve were taken as the characteristics of the selective amplifier circuit. Similar effects can be obtained even if they are approximate.

In addition, the number of selective amplification circuits is not limited to four;
It may be set to 8 for each. By doing this, you can perform more faithful ((listening, 〒 & level correction).

FIG. 2 shows another embodiment of the invention. In FIG. 2, the same reference numerals as in FIG. 1 indicate the same parts. 13 is an inverse selection amplifier circuit data memory, and 14 is a programmable inverse selection amplifier circuit. In this embodiment, the operation until the encoded data is decoded is the same as that shown in FIG. 1, so a description thereof will be omitted.

The inverse selection amplifier circuit data memory 13 stores in advance four sets of data defining characteristics opposite to each of the characteristics shown in FIG. is programmable reverse selection amplifier circuit 1
4 is output. Using this set of data, the programmable inverse selection amplifier circuit 14 is given a characteristic opposite to one of the characteristics shown in FIG. 11 within a certain period of time. Thus, the first
Similar to the embodiment shown in the figure, the signal whose amplitude has been compressed or expanded by the selective amplifier circuit is returned to the original input signal and output to the output terminal.

According to this embodiment, the processing circuit after decoding can be simpler than that in FIG. 1, and is economical.

FIG. 3 shows another embodiment of the invention. In FIG. 3, the same symbols as in FIG. 1 indicate the same parts. 15 is a selection filter circuit; 16-1 to 16-4 each have a flat frequency characteristic and a gain of 60.40, 20. OdB amplifier circuit, 17 attenuation data memory, 18 programmable attenuation circuit, 1
9 is a reverse selection filter circuit.

The selection filter circuit 15 has one of the frequency characteristics (indicated by ■) shown in FIG. 11, and has a gain of OdB from 1 to 5 KHz. This selection filter circuit 15 and each amplifier circuit 16-1 to 16-4 are combined to have the four characteristics shown in FIG. 12. For example, when combined with the amplifier circuit 16-1, the characteristics shown at the top of FIG. 12 are obtained. In other words, the combination of the selective filter circuit 15 and the amplifier circuits 16-1 to 16-4 has the same function as the selective amplifier circuits 2-1 to 2-4 shown in FIG. However, in Fig. 2, as shown in Fig. 11, the slope of the frequency-gain characteristic below I KFIz is different for each selection amplifier circuit, whereas in Fig. 3, the slope of the frequency-gain characteristic that determines the frequency-gain characteristic is different for each selection amplifier circuit. is common, so its slope (
The shapes (shapes) are the same, but the gains differ by 20 dB.

The attenuation data memory 17 stores -60 dB and -40 dB as attenuation data corresponding to the gains of the amplifier circuits 16-1 to 16-4.
It stores the values of dB, -20dB, and OdB.

This attenuation data memory 17 is addressed with the decoded selection signal to set the corresponding attenuation degree in the programmable attenuation circuit 18. Programmable attenuation circuit 18
attenuates the signal by the cent attenuation degree for a fixed time interval.

The inverse selection filter circuit 19 has a characteristic opposite to that of the selection filter circuit 15. The combination of this inverse selection filter circuit 19 and the programmable attenuation circuit 18 functions in the same way as the inverse selection augmentation circuit shown in FIG.

Since the signal processing operation is the same as that in FIG. 1, the explanation will be omitted.

Further, the connection heads of the inverse selection filter circuit and the programmable attenuation circuit may be reversed.

According to the embodiment shown in FIG. 3, the plurality of selective amplifier circuits in FIG. 81 and FIG. can.

In the above explanation, quantization/encoding was explained using linear PCM code 4 encoding as an example, but it is not limited to this, and ADPCM
, ATC, etc. may be used ().

Furthermore, although the case where quantization and encoding are performed using a fixed number of bits has been described, the number of quantization bits may be made variable depending on the selection of the selective amplifier circuit. This takes advantage of the fact that the lower the level of the signal, the coarser the required resolution. For example, when the output of the selective amplifier circuit 2-1 is selected, it is quantized to 5 bits, 2-2 is quantized to 6 bits, and 2-3.2-4 is quantized to 7 bits.

〔Effect of the invention〕

According to the present invention, it is possible to compress an audible sound signal to about 7/16 using the same encoding as a CD without degrading the perceptual quality. This has the effect of enabling long-term recording.

If more efficient encoding is used, compression of 7/16X4 to old is also possible.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing one embodiment of the invention, FIG. 2 is a block diagram showing another embodiment of the invention, FIG. 3 is a block diagram of another embodiment of the invention, and FIG. 4 is a block diagram showing another embodiment of the invention. The figure is a waveform diagram showing the pure tones of 100FIz and IKk at the same loudness level and their mixed signals, Figure 5 is a characteristic diagram showing the Fletcher Munson equal loudness curve, and Figure 6 is the waveform diagram of pure tones of 100FIz and IKk at the same loudness level and their mixed signals. Waveform diagrams showing pure tones and their mixed signals; Figure 7 is a characteristic diagram showing the relationship between loudness dynamic range and frequency; Figure 8
Figure 9 is a characteristic diagram showing the relationship between the sound intensity, L, and the ratio discrimination range and frequency. Figure 9 is a characteristic diagram showing the relationship between the sound intensity, L, and the sensory level. Figure 10 is the characteristic diagram of the selective amplification circuit. FIG. 11 is a characteristic diagram showing the characteristics of a linearly approximated selective amplifier circuit, and FIG. 12 is a characteristic diagram showing the characteristics of a combination of a selective filter circuit and an amplifier circuit. 2-1 to 2-4...Selection amplifier circuit, 3-1 to 3-4...Buffer memory circuit, 4-1 to 4
-4...Peak hold circuit, 7...Encoding circuit,
9...Decoding circuit, 11-1 to 11-4...
Reverse selection amplifier circuit. Representative Patent Attorney/Katsuo Togawa “-゛No. 40 Figure 5 FreqUenC7In C7CIeS Per
5eCO Had No. 60 7 Figure X Number (Hz) 8 Frequtn: i No. 60
II frequency, wave number CHL'1

Claims (1)

[Claims] 1. Encoding an audible sound signal into a digital signal and transmitting it;
In a system that decodes a transmitted digital signal and reproduces an audible signal, before encoding the audible signal,
An audible sound signal is divided into blocks at regular time intervals to obtain a block unit signal, and the amplitude of the block unit signal is increased to the maximum level using circuit means having the inverse characteristic of the Fletcher Munson equal loudness curve that adapts to the level. What is claimed is: 1. An audible signal transmission system characterized in that after decompression and decoding, amplitude compression is performed by circuit means having characteristics of a Fletschier-Manson equal loudness curve. 2. An audible signal transmission system according to claim 1, characterized in that information defining an equal loudness curve is transmitted as sub information at the same time as signal data.