JPH0955778A - Bandwidth widening device for sound signal - Google Patents

Abstract

PROBLEM TO BE SOLVED: To widen the bandwidth of a received narrow-band sound signal and improve reproduced sound quality by the bandwidth widening device for the sound signal which receives the narrow-band sound signal and widens its bandwidth. SOLUTION: This device is equipped with a fundamental frequency extraction part 1 which extracts the fundamental frequency of the narrow-band sound signal receives by a narrow-band sound receiver 4, a band expansion part 2 which finds a linear predicted residue signal by a linear predictive analysis of the marrow-band sound signal. shafts the linear predicted residue signal by an integral multiple of the fundamental frequency in the direction of a frequency axis, and performs linear predictive composition to output a band- expanded sound signal, and an addition part 3 which adds the received narrow- band sound signal and the sound signal having the band expanded by the band expansion part 2 together and applies a broad-band sound signal to a handset 5.

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 broadening device for receiving a narrow band voice signal on a telephone line or the like to broaden the band. In the production of human voice, for example, the vocal cords vibrate at a fundamental frequency (pitch frequency) of about 50 to 500 Hz to generate pulse-shaped sound waves (vocal cord waves), and when the vocal cords pass through the vocal tract resonance of the vocal tract. The component close to the frequency is emphasized and sent to the outside, and the band from the bass side of the bass to the treble side of the soprano is, for example, about 50 to 9000 Hz. Ordinary telephone line is 300-3400H
It has a band of z. Although the band of this telephone line is sufficient for identifying the content of the call, when it is applied to, for example, a multimedia communication system, it is necessary to bring it closer to a natural voice.

[0002]

2. Description of the Related Art A communication system using an ordinary telephone line is
For example, as shown in FIG. 9, one handset (speaker) 91, a narrow band voice transmitter 92, a transmission line 93 including a subscriber line, an exchange, etc., a narrow band voice receiver 94, and the other hand. And a handset (handset) 95 of
A narrow band voice transmitter 92,
An audio signal of 300 to 3400 Hz is transmitted according to the band by the transmission path 93 and the narrow band audio receiver 94.

FIG. 10 is an acoustic frequency characteristic curve diagram of a handset, showing an example of the characteristics of the handset of a normal telephone handset, which has a reproduction output band wider than 300 to 3400 Hz of the telephone line band. In addition, ISDN
The line has a wider band than an ordinary analog telephone line, for example, a voice signal of 50 to 7000 Hz is 64k.
It can be digitally encoded at bps and transmitted.

[0004]

In voice-only communication, since the contents of the call can be identified even in the communication system using the narrow band voice signal described above, no serious problem arises, but it is possible to have a conversation while seeing an image, In a multimedia communication system such as a teleconference, narrow band voice communication has a poor sense of presence and naturalness, and has a problem in serviceability. Wideband voice communication may be performed by applying the above-mentioned ISDN line, but since it is necessary to provide a voice encoder and a decoder for digitizing voice signals and the like, and the line usage fee is high, overall There is a problem that the communication cost becomes high. It is an object of the present invention to receive a narrow band voice signal and widen its band.

[0005]

An apparatus for broadening a voice signal according to the present invention will be described with reference to FIG. 1. (1) A fundamental frequency extraction unit 1 for detecting a fundamental frequency of a narrowband speech signal.
And a linear prediction residual signal of a narrow band speech signal is obtained, and the linear prediction residual signal is an integral multiple of the fundamental frequency. A band in which the frequency is shifted to either one or both of the high band side and the low band side and combined. The expansion unit 2 and the addition unit 3 that adds the audio signal expanded by the band expansion unit 2 and the narrowband audio signal and outputs a wideband audio signal are provided.

(2) Further, a linear prediction analysis unit and an inverse filter for obtaining a linear prediction residual signal by linear prediction analysis of the received narrow band speech signal, a fundamental frequency extraction unit for extracting a fundamental frequency of the narrow band speech signal, A shift unit for shifting the linear prediction residual signal in the frequency axis direction by an integer multiple of the fundamental frequency,
A linear prediction synthesis unit that performs linear prediction synthesis using the linear prediction residual signal shifted by the shift unit, and a wideband by adding the speech signal synthesized by this linear prediction synthesis unit and the received narrowband speech signal And an adder for converting the audio signal.

(3) A linear prediction analysis unit and an inverse filter for obtaining a linear prediction residual signal by linear prediction analysis of the received narrow band speech signal, a fundamental frequency extraction unit for extracting a fundamental frequency of the narrow band speech signal, A shift unit for shifting the linear prediction residual signal in the frequency axis direction by an integer multiple of the fundamental frequency,
A linear prediction synthesis unit that performs prediction synthesis as a wideband linear prediction residual signal obtained by adding the linear prediction residual signal and the linear prediction residual signal shifted by the shift unit.

(4) Further, the fundamental frequency extraction unit may be configured to extract the fundamental frequency of the narrow band speech signal from the linear prediction residual signal.

(5) Further, the fundamental frequency extraction unit can be configured to extract the fundamental frequency of the narrow band speech signal from the waveform of the received narrow band speech signal.

(6) Further, the fundamental frequency extraction unit can be configured to extract the fundamental frequency of the narrowband speech signal from the decoder of the narrowband speech signal encoded with the fundamental frequency as one of the parameters.

(7) The synthesized and output wideband audio signal is
A correction circuit for correcting the acoustic frequency characteristic of the handset can be provided.

[0012]

[Action]

(1) A narrowband voice signal is transmitted via a narrowband telephone line.
Received by the narrow band voice receiver 4, and the fundamental frequency extraction unit 1
The fundamental frequency of the narrow band speech signal is extracted by. The band extension unit 2 obtains a linear prediction residual signal of a narrow band speech signal,
This linear prediction residual signal is shifted by an integral multiple of the fundamental frequency to the high band side or low band side or both sides of the narrow band speech signal.
Then, the respective linear prediction residual signals are subjected to linear prediction synthesis, and the expanded band portion of the audio signal and the narrow band audio signal are added by the adder 3 and added to the handset (handset) 5 as a wide band audio signal. . Alternatively, the expanded portion of the linear prediction residual signal subjected to the frequency shift processing may be added to the narrowband residual signal to form a wideband residual signal, and the wideband residual signal may be linearly predicted and combined to form a wideband speech signal.

(2) Also, the received narrow band speech signal is predictively analyzed by a linear prediction analysis unit to obtain a linear prediction coefficient, and the narrow band speech signal is subjected to linear prediction residual through an inverse filter having an inverse characteristic of the linear prediction coefficient. The difference signal is obtained and added to the shift section, and the fundamental frequency of the narrowband speech signal is extracted by the fundamental frequency extraction section, and the linear prediction residual signal is shifted in the frequency axis direction by the integral multiple of this fundamental frequency in the shift section. To do. As a result, a band-extended linear prediction residual signal is obtained, and a speech signal is obtained by linear prediction synthesis, and a band-extended portion of this speech signal and a narrow-band speech signal are added to obtain a wide-band speech signal.

(3) Also, the received narrow band speech signal is predictively analyzed by a linear prediction analysis unit to obtain a linear prediction coefficient, and the narrow band speech signal is subjected to linear prediction residual through an inverse filter having an inverse characteristic of the linear prediction coefficient. The difference signal is obtained and added to the shift section, and the fundamental frequency of the narrowband speech signal is extracted by the fundamental frequency extraction section, and the linear prediction residual signal is shifted in the frequency axis direction by the integral multiple of this fundamental frequency in the shift section. To do. The band expansion part of this shifted linear prediction residual signal and the preceding linear prediction residual signal are added to form a wideband residual signal, and this residual signal is combined in the linear prediction combining unit to obtain the wideband speech. Signal.

(4) Since the voiced sound portion of the linear prediction residual signal has periodicity with the fundamental frequency as a period, the fundamental frequency of the narrowband speech signal is extracted by the fundamental frequency extraction unit. You can

(5) Further, the fundamental frequency extraction unit can extract the fundamental frequency of the narrow band speech signal by waveform processing of the narrow band speech signal.

(6) When a parameter including a fundamental frequency, amplitude, voiced / unvoiced judgment, etc. is used in the coding of a voice signal, the coded narrow band voice signal is received and decoded by a decoder. The fundamental frequency can be obtained as one of the time parameters. Therefore, using this fundamental frequency, band expansion can be performed by shifting the linear prediction residual signal.

(7) In addition, a wide band voice signal can be corrected by a correction circuit including a filter for correcting the acoustic frequency characteristic of the handset receiver, and the sound quality reproduced by the handset 5 can be further improved.

[0019]

1 is an explanatory view of a first embodiment of the present invention, in which 1 is a fundamental frequency extraction unit, 2 is a band expansion unit, 3 is an addition unit, 4 is a narrow band voice receiver, and 5 is a handset. (Handset). A narrow band voice signal is input to the narrow band voice receiver 4 via a normal telephone line (not shown). In this embodiment, a band broadening device is constituted by a fundamental frequency extraction unit 1, a band expansion unit 2, and an addition unit 3.

The fundamental frequency extraction unit 1 detects the fundamental frequency (pitch frequency) of the narrow band speech signal, and the band expansion unit 2
Calculates a linear prediction residual signal of a narrow band speech signal and frequency-shifts the linear prediction residual signal to either one or both of the high band side and the low band side by an integral multiple of the fundamental frequency and synthesizes them. That is, by performing linear prediction analysis and shifting the linear prediction residual signal to the high band side or low band side of the narrow band speech signal to perform linear prediction synthesis, it is possible to obtain a band-extended speech and The voice signal of the band-expanded portion is added to the narrow band voice signal in the section 3, and the wide band voice signal is added to the handset 5. Therefore, it is possible to make a call using a voice signal with improved quality.

For example, in the case of a narrow band audio signal of 300 to 3400 Hz, the band expanding unit 2 supplies 50 to 300.
By adding the audio signal of Hz and the audio signal of 3400 to 7000 Hz in the addition unit 3 via the band filter, a wide band audio signal of 50 to 7000 Hz can be formed.

FIG. 2 is an explanatory view of the second embodiment of the present invention, in which 10 is a band broadening device, 11 is a fundamental frequency extraction unit, and 1 is
2 is a band expansion unit, 13a is a high band processing unit, 13b is a low band processing unit, 14, 14a and 14b are band pass filters, and 15
Is an adder, 16 is a handset (receiver), 17 is a narrow band voice receiver, 18 is a transmission line such as a telephone line, 19 is a narrow band voice transmitter, and 20 is a handset (speaker).

The narrow band speech signal received by the narrow band speech receiver 17 is added to the band broadening device 10, and, for example, 300 to 3
Adder 15 via band pass filter 14 of 400 Hz
Add to Further, in the high frequency processing unit 13a, the basic frequency extraction unit 11 detects the basic frequency of the narrow band speech signal, and the band expansion unit 12 shifts the linear prediction residual signal to the high frequency side for linear prediction. It synthesizes, for example, 3400-
It is added to the adder 15 via the 7,000 Hz band pass filter 14a. Further, in the low frequency processing unit 13b, the basic frequency extraction unit 11 detects the basic frequency of the narrow band speech signal, and the band expansion unit 12 shifts the linear prediction residual signal to the low frequency side to perform linear prediction synthesis. For example, 50 to 30
It is added to the adder 15 through the 0 Hz band pass filter 14b.

Therefore, from the adder 15 to 50-7000H
The z broadband audio signal will be output and applied to the handset 16. That is, receiving a narrow band voice signal,
The band can be broadened by the band broadening device 10 and reproduced by the handset. Further, the high frequency processing unit 13a of this embodiment
And the fundamental frequency extraction unit 11 in the low frequency processing unit 13b
Can be made common because the fundamental frequency is extracted by the same processing.

FIG. 3 is an explanatory diagram of the third embodiment of the present invention, in which 21 is a linear prediction analysis unit, 22 is an inverse filter, 23 is a fundamental frequency extraction unit, 24 is a shift unit, and 25 is a linear prediction synthesis unit. , 26 is a band pass filter, 27 is a high pass filter, 28 is an adder, 29 is an AD converter (A / D), 30
Is a DA converter (D / A), and 31 is a handset (receiver).

300-340 received via telephone line
The 0 Hz analog narrow band speech signal is converted into a digital signal by the AD converter 29, the linear prediction analysis unit 21 obtains a linear prediction coefficient, and the narrow band speech signal is passed through an inverse filter 22 having an inverse characteristic of the linear prediction coefficient. In addition to the fundamental frequency extraction unit 23 and the shift unit 24, the fundamental frequency extraction unit 23 extracts the fundamental frequency (pitch frequency) of the narrowband speech signal, and the shift unit 24 performs linear prediction by an integral multiple of the fundamental frequency. The residual signal is shifted in the frequency axis direction.
The fundamental frequency extraction unit 23 can be configured to extract the fundamental frequency from the linear prediction residual signal, for example, by the already known autocorrelation method or the like.

Further, the linear prediction residual signal shifted by the shift unit 24 is synthesized by the linear prediction synthesis unit 25 into a band-extended voice signal, and the band pass filter 26 causes a voice signal of, for example, 50 to 300 Hz. Is added to the adder 28. The high-pass filter 27 is for passing an audio signal in a band of 300 Hz or higher, for example.
A broadband sound of 3400 Hz can be obtained. This is converted into an analog voice signal by the DA converter 30 and added to the handset 31. Therefore, it is possible to receive a narrow band audio signal of 300 to 3400 Hz and reproduce a wide band audio signal of 50 to 3400 Hz, for example.

FIG. 4 is an explanatory view of the main part of the shift part,
1, 42 are multipliers, 43 is a π / 2 phase shifter, and 44 is an adder. Assuming that the input signal is sin (ft) and the fundamental frequency of the narrow band speech signal is f ₀ , one multiplier 41 has cos (nf ₀ t) and the other multiplier 42 has sin.
Add (nf ₀ t). Therefore, the output signal of one multiplier 41 is sin (ft) · cos (nf ₀ t), and the output signal of the other multiplier 42 is sin (ft−π / 2) ·.
sin (nf ₀ t), and the output signal of the adder 44 is
sin (f-nf ₀ ) t. In this case, it is possible to shift nf ₀ , which is an integral multiple of the fundamental frequency f ₀ , to the low frequency side. Similarly, it is possible to shift to the high frequency side.

FIG. 5 is an explanatory view of the fourth embodiment of the present invention, in which 51 is a linear prediction analysis unit, 52 is an inverse filter, 53 is a fundamental frequency extraction unit, 54 is a shift unit, and 55 is a linear prediction synthesis unit. , 56 is a band pass filter, 57 is a high pass filter, 58 is an adder, 59 is an AD converter (A / D), 60
Is a DA converter (D / A).

Also in this embodiment, similar to the embodiment shown in FIG. 3, the narrow band voice signal received via the telephone line is AD-converted.
The converter 59 converts the signal into a digital signal, the fundamental frequency extraction unit 53 extracts the fundamental frequency of the narrow band speech signal, and the shift unit 54 extracts the linear prediction residual signal by an integral multiple of the fundamental frequency. The fundamental frequency extraction unit 53 extracts the fundamental frequency by waveform processing of the narrowband speech signal, and for example, has a configuration of extracting the fundamental frequency by the already-known modified correlation method or the like. can do. The operations of the linear prediction analysis unit 51, the inverse filter 52, the shift unit 54, the linear prediction synthesis unit 55, and the bandpass filter 56 are the same as those in the embodiment shown in FIG.

FIG. 6 is an explanatory diagram of a communication system according to a fifth embodiment of the present invention. 61 is a narrow band speech encoder, 62 is a narrow band speech transmitter, 63 is a transmission line including a telephone line, and 64 is a line. A narrow band voice receiver, 65 is a narrow band voice decoder, 66 is a wide band processing unit, and 67 and 68 are handsets. Narrowband speech encoder 61 enables handset 67 (speaker)
From the narrow band voice transmitter 62 to the transmission line 63.
Via the narrowband receiver 64, the narrowband speech decoder 65 decodes it, and the wideband processing unit 66
Add to

The narrow band speech decoder 65, as described above,
Since the basic frequency f ₀ is one of the parameters, the basic frequency f ₀ is extracted and added to the wide band processing unit 66. The band widening processing unit 66 shifts the linear prediction residual signal by an integer multiple of the fundamental frequency f ₀ to form a band-extended linear prediction residual signal, performs linear prediction synthesis of the linear prediction residual signal into a speech signal, and sets the handset 68. Add to (Handset).

FIG. 7 is an explanatory view of the fifth embodiment of the present invention, in which 71 is a linear prediction analysis unit, 72 is an inverse filter, 73 is a fundamental frequency extraction unit, 74 is a shift unit, and 75 is a linear prediction synthesis unit. , 76 is a band pass filter, 77 is a high pass filter, 78 is an adder, 79 is an AD converter (A / D), 80
Is a DA converter (D / A).

Since the fundamental frequency extraction unit 73 is added with the fundamental frequency f ₀ from the narrow band speech decoder 65 of FIG. 6,
The shift unit 74 uses it as the fundamental frequency of the narrow band audio signal.
Control the shift of the linear prediction residual signal in. Therefore, the basic frequency extraction unit 73 has a simple configuration for relaying the basic frequency f ₀ . In addition, the linear prediction analysis unit 71,
Inverse filter 72, shift unit 74, linear prediction synthesis unit 75,
The operation of the bandpass filter 76 will be described with reference to FIGS.
Since it is the same as the embodiment shown in FIG.

FIG. 8 is an explanatory diagram of a communication system according to a sixth embodiment of the present invention, in which 81 is a correction circuit, 82 is a narrow band voice transmitter, 83 is a transmission line including a telephone line, and 84 is a narrow band voice. Receiver, 85 is wideband processing unit, 86 is correction circuit, 87
Is a handset. This handset 8
7 has, for example, the acoustic frequency characteristic shown in FIG.

The wide band processing unit 85 corresponds to, for example, the wide band processing unit 66 of FIG. 6, and shifts the linear prediction residual signal of the narrow band speech signal in the frequency axis direction by an integer multiple of the fundamental frequency to perform linear prediction synthesis. The correction circuit 86 performs correction for emphasizing the low frequency band so as to further correct the acoustic frequency characteristic of the receiver. Thereby, the reproduced sound quality of the handset can be further improved. In this case, the wide band processing unit 85 shifts the linear prediction residual signal to the high frequency side to restore the high frequency side audio signal, and the correction circuit 86 emphasizes the high frequency band to improve the reproduced sound quality of the handset. You can also

Further, a correction circuit 81 may be provided on the transmission side of the narrow band audio signal to correct the acoustic frequency characteristic of the transmitter and further to correct the acoustic frequency characteristic of the receiver. By such correction and widening of the band of the audio signal by the wide band processing unit 85, the reproduced sound quality of the handset is further improved.

In the embodiments shown in FIGS. 3, 5 and 7, the linear prediction residual signal is shifted by the shift units 24, 54 and 74.
, And the linear prediction synthesis units 25, 55, and 75 perform linear prediction synthesis to generate a band-extended audio signal,
Although it is to be synthesized with a narrow band speech signal, the linear prediction residual signal and the linear prediction residual signal shifted in the frequency axis direction by an integral multiple of the fundamental frequency by the shift units 24, 54 and 74 are added to each other. A wideband residual signal can be formed, and the residual signal can be converted into a speech signal by linear prediction synthesis to widen the band. Further, by shifting the linear prediction residual signal to the high frequency side, the high frequency side of the narrow band speech signal can be restored to widen the band.

[0039]

As described above, according to the present invention, a narrow band speech signal is received and a wide band process is performed using the linear prediction residual signal and the fundamental frequency. There is an advantage that the voice quality can be improved so that it can be applied to a multimedia communication system.

[Brief description of drawings]

FIG. 1 is an explanatory diagram of a first embodiment of the present invention.

FIG. 2 is an explanatory diagram of a second embodiment of the present invention.

FIG. 3 is an explanatory diagram of a third embodiment of the present invention.

FIG. 4 is an explanatory view of a main part of a shift section.

FIG. 5 is an explanatory diagram of a fourth embodiment of the present invention.

FIG. 6 is an explanatory diagram of a communication system according to a fifth embodiment of this invention.

FIG. 7 is an explanatory diagram of a fifth embodiment of the present invention.

FIG. 8 is an explanatory diagram of a communication system according to a sixth embodiment of this invention.

FIG. 9 is an explanatory diagram of a conventional communication system.

FIG. 10 is an acoustic frequency characteristic curve diagram of the handset.

[Explanation of Codes] 1 fundamental frequency extraction unit 2 band expansion unit 3 addition unit 4 narrow band voice receiver 5 handset (handset)

Claims (7)

[Claims] 1. A voice signal band broadening device for receiving a narrow band voice signal to widen the band, comprising: a fundamental frequency extraction unit for detecting a fundamental frequency of the narrow band voice signal; and a linear shape of the narrow band voice signal. A band expansion unit that obtains a prediction residual signal, synthesizes the linear prediction residual signal by frequency-shifting an integer multiple of the fundamental frequency to either one or both of the high band side and the low band side, and the band. An apparatus for widening a voice signal, comprising: an addition unit that adds the voice signal extended by the extension unit and the narrowband voice signal to output a wideband voice signal. 2. A linear prediction analysis unit for obtaining a linear prediction residual signal by linear prediction analysis of the received narrow band speech signal and an inverse filter, and a fundamental frequency extraction unit for extracting a fundamental frequency of the narrow band speech signal. A shift unit that shifts the linear prediction residual signal in the frequency axis direction by an integer multiple of the fundamental frequency; and a linear prediction synthesis unit that performs linear prediction synthesis using the linear prediction residual signal shifted by the shift unit. 2. An adder for adding the extended band portion of the speech signal synthesized by the linear prediction synthesis unit and the received narrowband speech signal to obtain wideband speech. Audio signal broadening device. 3. A linear prediction analysis unit and an inverse filter that obtain a linear prediction residual signal by linear prediction analysis of the received narrowband speech signal, and a fundamental frequency extraction unit that extracts a fundamental frequency of the narrowband speech signal. A shift unit that shifts the linear prediction residual signal in the frequency axis direction by an integer multiple of the fundamental frequency, a wide band by adding the linear prediction residual signal and the linear prediction residual signal shifted by the shift unit. 2. The apparatus for widening a speech signal according to claim 1, further comprising: a linear prediction synthesis unit that performs predictive synthesis as the linear prediction residual signal. 4. The wide band of the speech signal according to claim 1, wherein the fundamental frequency extraction unit has a configuration for extracting a fundamental frequency of a narrow band speech signal from the linear prediction residual signal. Device. 5. The basic frequency extraction unit has a configuration for extracting a basic frequency of the narrowband speech signal from a waveform of the received narrowband speech signal.
Alternatively, the device for broadening the bandwidth of the audio signal according to 2 or 3. 6. The fundamental frequency extraction unit has a configuration for extracting the fundamental frequency of the narrowband speech from a decoder of the narrowband speech signal encoded with the fundamental frequency as one of the parameters. 4. The audio signal band broadening device according to claim 1, 2, or 3. 7. A wide band speech signal according to claim 1, further comprising a correction circuit for correcting the acoustic frequency characteristic of the receiver for the synthesized and output wide band speech signal. apparatus.