JPS63299419A - Transmission system for audible sound signal - Google Patents

Abstract

PURPOSE:To prevent sound quality in an audible sense from being deteriorated even when compressibility is raised, by performing amplitude correction by the reverse characteristic of a Fletcher-Munson's loudness-level contour on an audible sound signal before encoding, and performing the amplitude correction by the loudness-level contour again after decoding. CONSTITUTION:On an input signal, the reverse characteristic of the loudness- level contour regulated by the loudness level of the input signal is attached at a Fletcher-Munson's loudness-level contour reverse characteristic attaching part 4. And an amplitude-compressed input signal is, for example, PCM-encoded by an encoding part 6. Also, the loudness level is encoded at the encoding part 6, and is transmitted to a decoding part 8 as sub information. The decoding part 8 converts the encoded data of the input signal to the same electrical signal as the input signal by decoding. A reproducing signal is inputted to a Fletcher-Munson's loudness-level contour characteristic attaching part 9, and the characteristic of the loudness-level contour is applied. In such a way, it is possible to compress information without deteriorating the sound quality, and to improve transmission efficiency.

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 ratio without deteriorating 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 TV telephones. In order to answer this request,
PCM or even more efficient ADPCM for digital transmission of audible sounds.

Various encoding methods such as ATC are being researched and put into practical use.

For example, if we consider a well-known CD (compact disc) as an example of PCM that can obtain high-quality music signals, the amount of information is enormous at approximately 700 Kbit/5 (44.1 KHz sampling, 16-bit encoding). It is difficult to apply it to digital TV telephones and the like. For ATC, see IEE, Transactions on Audio, Speech, and Signals.
Processing, Volume, Ni Nis Nis P-25, Number 4, August 1977, No. 29
Pages 9 to 309 (I E E E Trans,
Acousties, 5peech, and, Sign
al Processing, Vol, ASSP-2
5, No, 4. August 1977).

The amount of information in ATC can be compressed to about 1/2 compared to the previous PCM without degrading the sound quality, but further compression will degrade the sound quality.

[Problem that the invention seeks to solve]

As explained above, in the above-mentioned conventional technology, when the compression ratio is increased, the sound quality deteriorates. This is because compression is performed without taking human auditory characteristics into account.

An object of the present invention is to take human auditory characteristics into consideration during encoding so that the sound quality does not deteriorate perceptually even when the compression rate is increased.

[Means for solving problems]

The above purpose is to apply the Flether-Muson (Flether-Mu
This is achieved by performing level correction using the inverse characteristics of the Fletcher Munson equal loudness curve, encoding, decoding, and then performing level correction using the Fletcher Munson equal loudness curve before listening.

[For production]

The level correction based on the inverse characteristic of Fletcher Munson's equal loudness curve operates to relatively lift audible sounds in the range of 1 to 5 KHz, for which human hearing sensitivity is high, and improve the S/N ratio in this band. Besides, 3.

Therefore, the energy of the audible sound is concentrated in this band, and the audible sound is appropriately encoded, for example, PCM encoded.

At this time, the number of quantization bits required for encoding is the minimum necessary to reproduce the signal amplitude in this band with high precision, and it is necessary to reproduce the large amplitude signal in the band with low auditory sensitivity as in the past. The number of quantization bits can be lower than the
It becomes possible to compress the amount of information. Furthermore, even with such compression, the perceptual sound quality does not deteriorate.

〔Example〕

Hereinafter, the present invention will be explained using figures.

Figure 3 shows Fletcher Munson's equal loudness curve. (Compiled by the Institute of Electronics and Communication Engineers, "Hearing and Speech," p. 104) The vertical axis is the sound pressure (amplitude), and the horizontal axis is the frequency. The numbers in the figure are the sound pressures, or loudness levels, of IKHz pure tones, and the numbers in parentheses are the loudness (sohn value).

For example, to perceive the same loudness (sound size) as a sound with IKHz and 40 dB sound pressure, 62 dB is required at 1007.
, 3 to 7 results in a sound with a sound pressure of 37 dB, and all pure tones accompanying this curve have the same loudness level of 40 phon. 10 to 90 phon, which is the normal hearing range for humans.
Considering this, the physical sound pressure, that is, the dynamic range of the amplitude level, is the same for all frequencies, but from the perspective of human hearing, or loudness, the dynamic range of the sound pressure is 100 Hz, and sounds with a sound pressure of 40 dB or less cannot be detected. Therefore, 9O-40=50 dB, and similarly for IK7, 9O-10=80 dB, and the dynamic range differs depending on the frequency.

Now, IKHz, 40dB (7) signal and 100Hz, 62
Consider a signal in FIG. 4 in which dB signals are mixed. As explained above, each signal is perceived by humans at the same loudness. For example, if the IKHz signal is
Assuming that 9 bits are required for high-precision (high-quality) encoding with M, the 100Hz signal amplitude is 62-40=22dB larger than the previous step to encode the mixed signal. A total of 13 bits (9 bits plus 22 ÷ 6 ÷ 4 bits) is required.

As explained above, in the PCM encoding used for conventional CDs, the necessary number of bits is determined solely from the dynamic range of the physical amplitude level, so 16 bits must be allocated to one sample. Ta.

On the other hand, a pre-emphasis method has been conventionally used as a method for improving the S/N of audio in audio analysis and audio recording and reproducing systems. For example, a single 6 dB 10 ct amplitude increase characteristic is applied to a voice signal, and this signal is analyzed or encoded. The audio signal is 6 dB due to the radiation characteristics of the mouth.
It has a falling characteristic of
) as the same sound pressure level as the high and low frequencies of
This prevents deterioration in analysis accuracy or sound quality (mainly S/N deterioration) due to a small number of bits.

However, with only the rise characteristic, when there is a signal with a high frequency (10KHz or higher) such as a music signal, the audible sensitivity will decrease even for high frequencies of 5KH7 or higher, as shown from the equal loudness curve in Figure 3. Inconvenience occurs because it is getting worse.

For example, the sound pressure of 10 KHz, which feels the same loudness as the sound of IKHz and 40 dB sound pressure, is 53 dB. The signal obtained by mixing these two signals is 6 dB/.

When multiplied by the increasing characteristic of ct, the current difference is 53-4
0=13dB is further increased by 20dB to become 33dB. (Figure 5) Therefore, as in the previous explanation, now IKI
(If 9 bits are required to encode the z signal with high precision using PCM, then the mixed signal requires 15 bits, which is 6 bits more than 33/6.

An embodiment of the present invention will be described below with reference to FIG. 1st
In the figure, 1 is an input terminal, 2 is a loudness level detection section, 3 is a Fletcher Munson loudness curve memory,
4 is a Fletcher Munson equivalent loudness curve inverse characteristic imparting section; 5 is a loudness level signal path; 6 is an encoding section; 7
8 is a transmission path, 8 is a decoding section, 9 is a Fletcher-Munson loudness curve characteristic imparting section, and 10 is an output terminal.

The input signal applied to the input terminal 1 is a spatial sound wave radiated from a sound source that is converted into an electrical signal by a microphone. The sound pressure at the microphone position is equivalent to the amplitude of the input signal. The loudness level of this signal is detected by the loudness level detection section 2. Loudness level (p
h o n value) is defined as the sound pressure level of a pure tone with a frequency of IKHz, and is usually measured with a sound level meter. In other words, the loudness level detection section 2 operates in the same way as a sound level meter, detects the loudness level of the input signal, and detects the loudness level of the input signal through the loudness level signal path 5.
Output to. The input signal is given an inverse characteristic of an equal loudness curve defined by the loudness level of the input signal in a Fletcher Munson equal loudness curve inverse characteristic imparting section 4 . The equal loudness curves are stored in advance in the Fletcher Munson equal loudness curve memory 3. The Fletcher Munson equal loudness curve inverse characteristic imparting unit 4 reads the corresponding equal loudness curve data from the Fletcher Munson equal loudness curve memory 3 based on the loudness level output from the loudness level detection unit 2, and also uses this data. In addition, the inverse characteristic is given to the input signal.

Currently, the loudness level of the input signal is 40ph.

If n, the equal loudness curve data shown by ■ in FIG.
The amplitude is normalized to a value representing the IKHz loudness level. For example, if the input signal is a pure tone of 1007 and has an amplitude equivalent to 62 dB, this amplitude is compressed by 22 dB and becomes an amplitude equivalent to 40 dB. Similarly, if the input signal is a pure tone of 10 KHz and has an amplitude equivalent to 53 dB, this amplitude is compressed by 13 dB and becomes an amplitude equivalent to 40 dB.

The input signal whose amplitude has been compressed in this manner is subjected to PCM encoding, for example, in the encoding section 6. Along with the encoding of this input signal, the loudness level is also encoded by the encoding section 6, and is transmitted to the decoding section 8 via the transmission path 7 together with the encoded data of the input signal as sub information.

The decoding unit 8 decodes the encoded data of the input signal and converts it into the same electrical signal as the input signal. The converted electrical signal is called a reproduced signal.

Loudness level encoded data as side information is also decoded at the same time, converted into a loudness level, and output to the loudness signal path 5.

The reproduced signal is input to the Fletcher-Munson loudness curve characteristic imparting section 9, which obtains equal-loudness curve data corresponding to the decoded loudness level from the Fletcher-Munson loudness curve memory 3, and converts it into a reproduced signal based on this. On the other hand, the characteristic of this equal loudness curve (broken line in FIG. 6) is multiplied. For example, if the reproduced signal is a pure tone of 1007 and has an amplitude equivalent to 40 dB, it is expanded by 22 dB and has an amplitude equivalent to 64 dB, which is output to the output terminal 10.

Discrimination threshold (D, L,
), in the range of sound pressure that humans can perceive (from the minimum audible value of 0 phon to the maximum audible value of 120 phon), the number of steps that can be analyzed, that is, the number of steps that can be heard to tell the difference, is at most 280 steps. be. (Formula (2,
13)) In other words, this number of steps is 8 bits (21'=
256) can be expressed in 9 bits (29=512). In other words, for human hearing, it is sufficient to express the sound pressure of an audible sound in 8 to 9 bits, and finer resolution (representing the sound pressure in 9 bits or more) is not necessary. When sound pressure is converted into an electrical signal, it is equivalent to its amplitude. In other words, the signal amplitude can be expressed using 8 to 9 bits.

The above explanation of resolution refers to the loudness level, that is, the perceived loudness of the sound. If we consider only the physical sound pressure level, we can see from the example of PCM encoding used for CDs that if a pure tone of IKHz can be expressed in high quality with a resolution of 9 bits, then various frequencies can be expressed. For music signals that can be considered to be a mixture of pure tones, it is necessary to express the signal amplitude with a resolution of about 16 bits, which is an increase of about 7 bits. That is, without the present invention, when pure tones of different frequencies at the same loudness level are heard at the same time, they will be perceived as having the same quality only after the signal is expressed with 16-bit accuracy.

In the embodiment shown in FIG. 1, the Fletcher-Munson equivalent loudness curve inverse characteristic imparting section 4 converts the amplitude of the input signal from a scale proportional to the physical sound pressure level to a scale proportional to the psychological loudness sound pressure level. have. As a result, regardless of the frequency, it is sufficient to express the amplitude with the above-mentioned 9-bit resolution that can be analyzed by humans, and this does not result in any audible quality degradation.

In the above explanation, PCM was considered as the encoding method for the sake of simplicity, but it is clear that the same applies to other encoding methods such as ADPCM and ATC.

Further, in the embodiment, the encoded data is transmitted over a transmission path, but it is also obvious that the encoded data may be recorded on another recording medium, such as an optical disk, and then reproduced and decoded again.

Further, it is not necessary to continuously store equal loudness curves at all loudness levels in the Fletcher Munson equal loudness memory 3, but discretely, for example, equal loudness curves at all loudness levels.
0, 20, 30. ...Data for each 90 phon and 10 phon may be stored. At this time, the loudness level detection unit 2 converts the actual phon value into this discrete Ph
It is rounded to an on value and output, and this discrete phon value is encoded and transmitted.

Also, one standard loudness level, for example 40pho
n and its equal loudness curves are determined, and the difference between each equal loudness curve and the standard equal loudness curve is stored in the Fletcher Munson equal loudness memory 3, and the difference from the standard loudness level is encoded as the loudness level. May be transmitted.

FIG. 2 shows another embodiment of the invention. In FIG. 2, the same reference numerals as in FIG. 1 indicate the same parts. 11 is a changeover switch;
12 is a buffer memory. The changeover switch 11 and the buffer memory 12 constitute a two-sided buffer, which functions to divide the input signal into certain time units. The input signal is divided into blocks at certain time intervals, and the average loudness level of the input signal within the block is detected by a loudness level detection section. The subsequent operations are the same as those shown in FIG. 1 and will therefore be omitted.

In this embodiment, only one loudness level is transmitted as sub information for each block time interval, and the amount of loudness level information to be transmitted can be reduced.

〔Effect of the invention〕

According to the present invention, when considering PCM encoding, there is no audible quality deterioration even if information is compressed to 9/16 compared to the conventional method. A
When considering encoding such as DPCM and ATC, it is possible to compress information to 1/2 compared to PCM encoding even without using the present invention. /16 = 9/32, it becomes possible to compress information without deteriorating the sound quality, and the transmission efficiency can be significantly improved.

[Brief explanation of drawings]

Fig. 1 is a block diagram of one embodiment of the present invention, Fig. 2 is a block diagram of another embodiment, Fig. 3 is a Fletcher Munson equal loudness curve, and Fig. 4 is a diagram of the same loudness level of 100 Hz and 100 Hz. An explanatory diagram showing pure tones of IKI (z) and their mixed signals. Figure 5 is an explanatory diagram showing pure tones of IKHz and 10 KHz at the same loudness level and their mixed signals. Figure 6 is an illustration of equal loudness curves. 3.Fletcher-Munson loudness curve memo 1c4...Fletcher-Munson loudness curve inverse characteristic imparting part, 9.Fletcher-Munson loudness curve characteristic imparting part.

Claims (1)

[Claims] 1. Encode the audible sound signal into a digital signal and transmit it,
In a system that decodes the transmission signal and reproduces an audible signal, the audible signal is subjected to amplitude correction based on the inverse characteristic of Fletcher Munson's equal loudness curve before encoding, and the amplitude of the equal loudness curve is corrected after decoding. An audible signal transmission system characterized by performing amplitude correction based on characteristics.