JPH08321792A - Audio signal band compressed transmission method - Google Patents

Abstract

PURPOSE: To keep high speech quality by sufficiently suppressing deterioration in articulation accompanying band compression. CONSTITUTION: The method is provided with a 1st conversion section consisting of an LPF 2, a multiplier circuit 5, a sine wave signal source 8, and an LPF 11, with a 2nd conversion section consisting of an LPF 3, a multiplier circuit 6, a sine wave signal source 9, and an LPF 12, and with a 3rd conversion section consisting of an LPF 4, a multiplier circuit 7, a sine wave signal source 10, and an LPF 13. The low frequency components of 300Hz to 900Hz of an original voice signal fed to an input terminal 1 are converted into 0Hz to 600Hz, medium frequency components of 1300Hz to 1800Hz are converted into 600Hz to 1100Hz, and high frequency components of 2200Hz to 2600Hz are converted into 1100Hz to 1500Hz and sent.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a voice signal band compression method, and more particularly to a voice signal band compression transmission method suitable for use in a radio transmission system.

[0002]

2. Description of the Related Art In recent years, the utilization rate of wireless transmission lines has been increasing, but on the other hand, the radio frequency band is a finite resource. Therefore, compression of the occupied frequency band is not only for cost reduction, There is also a strong demand for effective use of resources.

Regarding the transmission of a voice signal, the frequency bandwidth of the voice signal generally extends over several kilohertz although there are individual differences. Therefore, the transmission of this also involves transmission of a frequency bandwidth of several kilohertz. Need a system,
Here, if the occupied frequency bandwidth can be compressed without impairing the intelligibility required for voice information transmission, the cost required for the transmission system can be reduced and the wireless transmission system can be used more effectively. .

The compression of the occupied frequency band is not limited to the wireless transmission line, and is a constant subject in the technical field concerned regardless of the form of signal transmission such as wired wireless.

Therefore, various audio signal band compression techniques have been proposed in the past.
According to Japanese Patent Laid-Open No. 344027, a signal having a part of the bandwidth within a band is extracted from an audio signal having a predetermined frequency band, and the extracted part is transmitted by converting the frequency to a low frequency side and transmitting the result. We have proposed a so-called voice band shift method capable of obtaining band compression, and an example of applying it to a communication device of a single frequency simultaneous transmission / reception system is described.

[0006]

The above-mentioned prior art cannot be said to give sufficient consideration to the deterioration of intelligibility due to band compression, and there is a problem that communication quality remains unsatisfactory. The object of the present invention is to sufficiently suppress the deterioration of intelligibility due to band compression, and to easily maintain a high communication quality even when there is a severe restriction on an occupied band such as a communication device of a single frequency simultaneous transmission / reception system. An object of the present invention is to provide a voice signal band compression transmission method that is made possible.

[0007]

The above-mentioned object is to extract a signal having a partial bandwidth within the band from a voice signal having a predetermined frequency band, and fold the extracted part back to the low frequency side for frequency conversion. In the audio signal band compression transmission method in which the transmission band is compressed by transmitting the signal in the above-mentioned manner, the signal of the above-mentioned part of the bandwidth is the low-frequency part, the mid-frequency part, and the high-frequency part in the frequency band. This is achieved by having the signal extracted from the part.

[0008]

In the case of Japanese, the formants that characterize the phoneme of voiced speech are distributed between 300 Hz and 900 Hz in the first formant and 1300 Hz to 180 in the second formant.
0 Hz, and the third formant is 1800 Hz to 250
It is distributed at 0 Hz.

In the present invention, the frequency band portion extracted and transmitted from the audio signal frequency band is the low frequency band portion, the middle frequency band portion, and the high frequency band portion. As a result, the low frequency band portion is The first formant distributed in the range of 300 Hz to 900 Hz is included, and 160 is included in the mid-range part and the high-range part.
A second formant and a third formant distributed between 0 Hz and 2500 Hz can be included.

Therefore, it is possible to exclude the first formant, the second formant, and the third formant in the part that is missing due to the partial extraction for band compression, and as a result, the clarity is improved. It is possible to sufficiently obtain the storage of each formant required for holding the call, and it is possible to minimize deterioration of the call quality due to band compression.

[0011]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A voice signal band compression transmission method according to the present invention will be described in detail below with reference to embodiments shown in the drawings. FIG.
In the embodiment of the present invention, only the conversion unit A for band compression on the transmission side and the decompression unit B on the reception side are shown.

In FIG. 1, first, the conversion unit A includes an input terminal 1 to which an audio signal to be band-compressed and transmitted is supplied,
900Hz which extracts the low frequency part from this audio signal
Low pass filter 2 of the same, and also extracts the mid-range part 18
It is provided with a low pass filter 3 of 00 Hz and a low pass filter 4 of 2600 Hz which also extracts a high band portion.

Next, the multiplication circuits 5 to 7 for folding back the respective components extracted by the low-pass filters 2 to 4 and the folding circuits for transforming the low-pass components to the frequency components of 900 Hz to 0 Hz are necessary. A sine wave signal source 8 that supplies a 900 Hz sine wave signal, and 2 in the middle range.
A 2400 Hz sine wave signal source 9 necessary for aliasing conversion to a frequency component of 400 Hz to 600 Hz, and a 3700 Hz sine wave signal source 10 also necessary for aliasing conversion of a high frequency part to a frequency component of 3400 Hz to 1100 Hz. Is equipped with.

Further, the multiplication circuit 5 causes 900 Hz to 0H.
60 from the low-frequency part folded back to the frequency component of z
Low-pass filter 11 for removing unnecessary parts above 0 Hz
2400 Hz to 600 Hz by the multiplication circuit 6
From the mid-range part that is folded back to the frequency component of
Low-pass filter 1 that removes unnecessary parts above 0 Hz
2, 3700Hz to 1100H by the multiplication circuit 7
15 from the high-frequency part folded back to the z frequency component
Low-pass filter 1 that removes unnecessary parts above 00Hz
3, an adder circuit 14 for synthesizing the outputs of the respective low-pass filters 11 to 13, and an output terminal 15 for extracting a band-compressed signal.

On the other hand, also in FIG. 1, the restoration unit B is
An input terminal 16 to which a band-compressed signal to be restored to a normal audio signal is supplied from the transmitting side, and a low-frequency part of this band-compressed signal is separated and taken out 600H
low-pass filter 17 for z, low-pass filter 18 for 1100 Hz, which also separates and extracts only the mid-range part, and 1500H, which similarly separates and extracts only high-frequency part
and a low pass filter 19 of z.

Next, the multiplication circuits 20 to 22 for folding back the respective components separated by the low-pass filters 17 to 18 to return them to the original frequency band, and the low-pass parts to the original 300 Hz to 900 Hz. And a sine wave signal source 23 that supplies a sine wave signal of 900 Hz necessary for re-folding conversion into frequency components of
A sine wave signal source 24 of 2400 Hz necessary for re-folding conversion into a frequency component of 00 Hz to 1800 Hz, and a sine wave signal source of 3700 Hz necessary for re-folding conversion of a high frequency part into a frequency component of 2200 Hz to 2600 Hz. 25 and.

Further, a low-pass filter 261 for removing unnecessary portions of 900 Hz or higher from the low-pass portion re-folded and converted by the multiplication circuit 20, and a middle-pass portion 1800 Hz re-folded and converted by the multiplication circuit 21 as well.
The low-pass filter 27 that removes the above unnecessary portions, the low-pass filter 28 that removes unnecessary portions at 2600 Hz or higher from the high-pass portion that is again folded and converted by the multiplication circuit 22, and these low-pass filters 26 to 28 are also included. It is provided with an adder circuit 29 for synthesizing outputs, and an output terminal 30 for taking out the audio signal restored thereto.

Next, the operation of the embodiment shown in FIG. 1 will be described. In the converter A on the transmission side, the input terminal 1 has
A voice signal in the same frequency band as that of an ordinary telephone is supplied, and therefore, the upper limit of the frequency band is approximately 3000 Hz. It should be noted that the lower limit of the frequency band is about several hundreds Hz, but it is described here that the frequency band extends to nearly 0 Hz, and therefore the bandwidth is also 3000 Hz.

The audio signal input to the input terminal 1 includes three low-pass filters (hereinafter simply referred to as filters) 2 to 4
As shown in FIG.
As shown in FIG. 4, only the portion of the audio signal below 900 Hz is taken out, then, as shown in FIG. 4, the portion of 1800 Hz and below is extracted to the output of the filter 3, and the output of the filter 4 is obtained. , As shown in FIG. 6, 2600H
The part below z is taken out.

The arrows in FIGS. 2, 4, and 6 and FIGS. 3, 5, and 7 to 9 described later indicate the frequency of the audio signal supplied to the input terminal 1, that is, the original signal. Indicates the direction in which is higher.

Next, the audio signals divided into these three kinds of frequency bands are respectively inputted to the multiplication circuits 5 to 7 and are supplied from the sine wave signal source 8 with a sine wave signal of 900 Hz or a sine wave signal. 2400 sourced from source 9
The sine wave signal of Hz or the sine wave signal of 3700 Hz supplied from the sine wave signal source 10 is respectively multiplied.

Therefore, from these multiplying circuits 5 to 7, the sum signal component of the respective frequencies and the difference signal component are obtained. Of these, the difference signal components are shown in FIG. As shown in FIG. 3, FIG. 5, and FIG. 7, the frequency of the original signal is folded and converted into a frequency-converted signal.

Therefore, the outputs of these multiplication circuits 5-7 are
The signals are input to the filters 11 to 13, respectively, and the filter 11 extracts only the components of 600 Hz or less,
If only the component of 1100 Hz or less is taken out, and if the component of 1500 Hz or less is taken out by the filter 13, the signal component of the frequency band shown by the thick solid lines in FIGS. 3, 5, and 7 is extracted at each output. To be done.

Therefore, looking at each of the signal components in FIGS. 3 to 7, first, in the output of the filter 11 shown in FIG. 3, the signal component in the frequency band of 300 Hz to 900 Hz of the original audio signal, that is, the low frequency band is generated. As shown by the arrow, the part is folded back to the low frequency side and the frequency is converted to 0H.
It can be seen that the state is shifted from the vicinity of z to the frequency band of 600 Hz.

Similarly, in the output of the filter 12 shown in FIG. 5, the signal component in the frequency band of 1300 Hz to 1800 Hz of the original audio signal, that is, the mid-range part, is folded back to the low frequency side as indicated by the arrow and the frequency is converted. In the output of the filter 13 shown in FIG. 7, a signal component in the frequency band of 1300 Hz to 1800 Hz of the original audio signal, that is, a high frequency band, is generated. It can be seen that the portion is folded and frequency-converted to the low frequency side as shown by the arrow, and is shifted to the frequency band of 1100 Hz to 1500 Hz.

Since the outputs of the filters 11 to 13 are added by the adder circuit 14, the output, that is, the output terminal 14 is eventually changed to that shown in FIG.
~ A signal in which the signal components of Fig. 7 are combined is obtained.

Then, as is apparent from FIG. 8, the signal appearing at the output terminal 14 has an upper frequency limit of 1
It can be seen that the signal has a bandwidth of about 1500 Hz at 500 Hz. At this time, since the audio signal input to the input terminal 1 has a bandwidth of 3000 Hz as shown in the figure, the output terminal 14 is eventually connected. , A band-compressed signal of 1/2, that is, a band-compressed signal is obtained.

Next, the band-compressed signal obtained at the output terminal 14 is transmitted to the receiving side by any known transmission method, supplied to the input terminal 16 of the decompression unit B, and decompressed to the original audio signal. To be done. Therefore, first, the input band terminal 16
The band-compressed signal supplied to the three filters 17, 1
Separate by 8 and 19. Next, the band compression signals divided into these three types of frequency bands are respectively supplied to the multiplication circuit 20.
To 22 and is supplied from the sine wave signal source 23, the 900 Hz sine wave signal, or the 2400 Hz sine wave signal supplied from the sine wave signal source 24, or the sine wave signal source 25. Each is multiplied by a 3700 Hz sine wave signal.

Therefore, from these multiplication circuits 20 to 22, the sum signal component of the respective frequencies and the difference signal component are obtained. Of these, the difference signal components are shown in FIG. The frequency of the band-compressed signal shown in each part of 8 is folded back again to obtain a frequency-converted signal.

Therefore, the outputs of these multiplication circuits 20 to 22 are input to the filters 26 to 28, respectively, and the filter 2
6 takes out only the component of 900 Hz or less, the filter 27 takes out only the component of 1800 Hz or less, and the filter 28 takes out only the component of 2600 Hz or less. Then, as shown in FIG. Thus, a restored audio signal is obtained, which is composed of a low-frequency part whose frequency direction is the same as that of the audio signal, a mid-range part, and a signal in each band of the high-frequency part.

Therefore, if the restored voice signal obtained at the output terminal 30 of the restorer B is set to a predetermined level and supplied to a speaker or an earphone via a predetermined drive circuit, the voice based on the original voice signal is reproduced. be able to.

By the way, according to this embodiment, as is clear from FIG. 9, the signal components of the low-frequency part, the mid-frequency part, and the high-frequency part of the band of the original audio signal are transmitted. However, as shown in the figure, there are many missing parts that are not transmitted.

However, in the voice band shift method, since band compression is obtained by intentionally providing such a missing portion, how to make a call while increasing the number of missing portions as much as possible. The ability to suppress the deterioration of quality is a factor that determines its performance.

Therefore, the performance obtained by the above embodiment will be described below. First, according to the present invention, the reason why the deterioration of speech quality due to band compression can be suppressed to a minimum is that the part to be extracted from the frequency band of the voice signal to be band-compressed and transmitted is a low band part. This is because it was set in three parts, the high frequency part and the high frequency part.

The formants that characterize the phoneme of voiced speech are
In Japanese, the first formant is 300Hz-900H
It is distributed in z and the second formant is 1300 Hz to 1
800 Hz, and the third formant is 1800 Hz to 2
It is distributed at 500 Hz.

In the present invention, the frequency band portion extracted from the audio signal frequency band and transmitted is the low frequency band portion, the middle frequency band portion and the high frequency band portion. A first formant distributed in the range of 300 Hz to 900 Hz is included in the midrange part, and 16 is included in the midrange part.
Includes a second formant distributed from 00Hz to 1800Hz, and 1800Hz to 2500 in the high range.
A third formant distributed in Hz can be included.

Therefore, it is possible to prevent the first formant, the second formant, and the third formant from being included in the part that is missing due to the partial extraction for band compression. It is possible to sufficiently obtain the storage of each formant required for holding the call, and it is possible to minimize the deterioration of the call quality due to the band compression.

This will be described more specifically with reference to the result of simulation of the embodiment shown in FIG. FIG.
There are seven types of filters used in the embodiment of the present invention when classified by cutoff frequency. As these filters, a 15th-order inverse Chebyshev filter having no ripple in the pass band and having a sufficiently steep transition band is used. Used, their minimum attenuation in the stopband is 40db and their characteristics are
It is as shown in FIGS.

The cutoff frequencies are exactly 600 Hz, 900 Hz, 1100, respectively.
Hz, 1500Hz, 1800Hz, 2500Hz, and 2600Hz, when folding frequency conversion, overlapping occurs between different frequency bands and accurate restoration cannot be performed, so 95% of each frequency
A filter having a cutoff frequency of about 97% is used.

Therefore, when a signal having the frequency characteristic shown in FIG. 17 is input to the input terminal 1 of the conversion unit A using these filters, the output terminal 15 is band-compressed with the frequency characteristic shown in FIG. A signal is obtained, and the output terminal 3 of the restoration unit B
In 0, the audio signal having the frequency characteristic shown in FIG. 19 could be obtained.

Next, this example was evaluated by a sensory test. First, in this evaluation, a signal having the characteristics shown in FIG. 20 was used as the original signal, and the following five transmission systems were compared.

Signal only passed through a 3000 Hz low pass filter. 300Hz ~ 1800H by voice band shift method
Signal shifted to z. 300Hz-900Hz by voice band shift method
And 1600 Hz to 2500 Hz divided into two parts, and the folding frequency conversion is performed so that the bandwidth becomes 1500 Hz,
The restored signal. 300Hz-900Hz by voice band shift method
And 1300 Hz to 1800 Hz, and 2200 Hz to
Divided into 3 parts of 2600Hz, bandwidth is 1500Hz
The restored signal is the frequency converted so that it becomes. Signal only passed through a 1500 Hz low pass filter.

21 to 25 show the frequency characteristics of each of the above signals. Here, corresponds to the embodiment of the present invention.

Then, these signals were reproduced as sound from a speaker and evaluated. Five evaluators A to E participated in the evaluation, and the intelligibility was evaluated. That is, all consonants that appear in standard Japanese, 95 syllables with 5 vowels connected, and 5 single vowels "A", "I", "U", "E", " According to the equality syllable table consisting of "o", 5 kinds of voice data are created from 100 syllables uttered by one man and 195 single-note nonsense syllables, and any one of the above-mentioned processes It was played on a speaker from and was evaluated based on how much of it was understood.

FIG. 26 shows the evaluation results, and it is natural that the evaluation is high when the bandwidth is 3000 Hz.
Among the 1500 Hz bandwidth signals of, the embodiment of the present invention has the highest evaluation.

Therefore, according to the embodiment of the present invention, although the frequency band is compressed to 1/2, the deterioration of the intelligibility is sufficiently suppressed and the high communication quality can be maintained.
As a result, according to the present invention, it is possible to easily apply even if there is a severe restriction on the occupied bandwidth, such as a single-frequency simultaneous transmission / reception type communication device, and it is possible to effectively use the communication channel while maintaining high communication quality. Can be achieved.

[0047]

According to the present invention, the portion extracted and transmitted from the frequency band of the original audio signal is set to the low band portion, the middle band portion and the high band portion within the band. Due to the partial extraction for band compression, it is possible to exclude the first formant, the second formant, and the third formant from the part that is missing from the original speech signal. It is possible to sufficiently obtain the storage of each formant required for holding the call, and it is possible to minimize deterioration of the call quality due to band compression.

[Brief description of drawings]

FIG. 1 is a block diagram showing a system to which an embodiment of a voice signal band compression transmission method according to the present invention is applied.

FIG. 2 is a characteristic diagram for explaining the operation of one embodiment of the present invention.

FIG. 3 is a characteristic diagram for explaining the operation of the embodiment of the present invention.

FIG. 4 is a characteristic diagram for explaining the operation of the embodiment of the present invention.

FIG. 5 is a characteristic diagram for explaining the operation of the embodiment of the present invention.

FIG. 6 is a characteristic diagram for explaining the operation of the embodiment of the present invention.

FIG. 7 is a characteristic diagram for explaining the operation of the embodiment of the present invention.

FIG. 8 is a characteristic diagram for explaining the operation of the embodiment of the present invention.

FIG. 9 is a characteristic diagram for explaining the operation of the embodiment of the present invention.

FIG. 10 is a characteristic diagram of a low pass filter used in an embodiment of the present invention.

FIG. 11 is a characteristic diagram of a low pass filter used in an embodiment of the present invention.

FIG. 12 is a characteristic diagram of a low pass filter used in an embodiment of the present invention.

FIG. 13 is a characteristic diagram of a low pass filter used in an embodiment of the present invention.

FIG. 14 is a characteristic diagram of a low pass filter used in an embodiment of the present invention.

FIG. 15 is a characteristic diagram of a low pass filter used in an embodiment of the present invention.

FIG. 16 is a characteristic diagram of a low pass filter used in an embodiment of the present invention.

FIG. 17 is a characteristic diagram of an original audio signal used in an example of the present invention.

FIG. 18 is a characteristic diagram of a band-compressed signal obtained in an example of the present invention.

FIG. 19 is a characteristic diagram of an audio signal restored in an embodiment of the present invention.

FIG. 20 is a characteristic diagram of an original speech signal used for intelligibility evaluation.

FIG. 21 is a characteristic diagram showing a first example of a voice signal used for intelligibility evaluation.

FIG. 22 is a characteristic diagram showing a second example of an audio signal used for intelligibility evaluation.

FIG. 23 is a characteristic diagram showing a third example of a voice signal used for clarity evaluation.

FIG. 24 is a characteristic diagram showing a fourth example of an audio signal used for intelligibility evaluation.

FIG. 25 is a characteristic diagram showing a fifth example of an audio signal used for intelligibility evaluation.

FIG. 26 is an explanatory diagram showing a result of an intelligibility evaluation.

[Explanation of symbols]

 A conversion unit on the transmission side B restoration unit on the reception side 1 Input terminals of the conversion unit 2, 3, 4 Low pass filter (LPF) 5, 6, 7 Multiplier circuit 8, 9, 10 Sine wave signal source 11, 12, 13 Low-pass filter (LPF) 14 Adder circuit 15 Output terminal of conversion unit 16 Input terminal of restoration unit 17, 18, 19 Low-pass filter (LPF) 20, 21, 22 Multiplier circuit 23, 24, 25 Sine Wave signal source 26, 27, 28 Low pass filter (LPF) 29 Adder circuit 30 Output terminal of restoration unit

Claims (2)

[Claims] 1. A transmission band by extracting a signal of a part of the bandwidth from a voice signal having a predetermined frequency band, folding back the extracted part to a low frequency side and transmitting the result. In the audio signal band compression transmission method that enables the compression of the above, the signal of the part of the bandwidth is a signal extracted from the low band part, the middle band part, and the high band part in the frequency band. A voice signal band compression transmission method characterized in that 2. The invention according to claim 1, wherein the predetermined frequency band is 300 Hz to 2600 Hz, the low frequency range is 300 Hz to 900 Hz, the middle frequency range is 1300 Hz to 1800 Hz, and the high frequency range is high. The range part is in the range of 2200 Hz to 2600 Hz, the low range part is folded and frequency-converted into a frequency band of 0 Hz to 600 Hz and then transmitted, and the middle part is folded and converted into a frequency band of 600 Hz to 1100 Hz. The audio signal band compression transmission method is characterized in that the high-frequency part is transmitted after being folded and frequency-converted into a frequency band of 1100 Hz to 1500 Hz and then transmitted.