JPH064087A - Speech encoding device - Google Patents

Abstract

PURPOSE:To reduce a feeling of incongruity to a reproduced sound at the time of switching between a voiced sound section and a voiceless sound section of the speech encoding device. CONSTITUTION:The speech encoding device, which has a voiceless sound detection part 1 detecting the voiceless section of an input speech signal and outputting voiceless sound information and an encoding part 2 encoding and outputting the voiceless sound information and inputting speech signal on the transmission side and a decoding part 3 decoding the received voiceless sound information and inputting speech signal and outputting a reproduced speech signal and the voiceless sound information and a noise insertion part 4 inserting a noise of specific level into the voiceless sound section of the reproduced sound signal according to the voiceless sound information outputted by the decoding part and outputting the resulting signal on the reception side, is provided with a time delay part 5 which is connected between the voiceless sound detection part 1 and encoding part 2 on the transmission side and outputs the voiceless sound information with specific time delay behind the point of time when the voiceless sound detection part decides the voiceless sound section; and the noise insertion part 4 on the reception side inserts the noise of specific level into the voiceless sound section of the reproduced speech signal and the certain section right before the voiceless sound section according to the voiceless sound information outputted by the decoding part and outputs the resulting signal.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an improvement of a speech coder. At this time, there is a demand for a voice encoding device that reduces a feeling of strangeness in reproduced sound that occurs when a sound section and a silent section are switched.

[0002]

2. Description of the Related Art FIG. 6 is a block diagram showing the structure of a conventional speech coder. In a voice encoding device that transmits a plurality of voice data or other data by multiplexing,
In order to improve the transmission efficiency of the transmission path, the silent section that occupies most of the voice data during a call is sequentially detected, and the transmission path is set to other voice data in the voice section or other data for which transmission is requested. A released data compression transmission method is adopted.

FIG. 6 shows an example of the configuration of a speech coder when the silence detection method is adopted. On the encoder side shown in FIG. 3A, the silence detection unit 1 determines that the voice level of the input voice data is less than a predetermined threshold for a certain period of time or more and determines that it is a silence section, and the determination result (Less than,
“Silence information”) is added to the encoding unit 2 together with the voice data. In the encoding unit 2, the silence information is targeted for encoding for the silent section, and the voice data is compressed and encoded by a predetermined compression encoding method for the voiced section,
Send to the transmission path.

On the other hand, in the voice encoding device facing each other through the transmission path, the decoding unit 3 shown in FIG. 1B reproduces the voice data from the compression encoded data and extracts the silence information. , Noise adding section 4 ′. Noise insertion unit 4 '
Then, a predetermined noise signal is inserted and output in the silent section of the reproduced voice data. The sound reproduced in this way is prevented from becoming completely silent.

[0005]

However, in the configuration of the above-mentioned device, the presence / absence of voice is discriminated by the voice level, and in order to perform the silence processing immediately after the level becomes lower than the threshold value, There is a problem in that the decoder causes an uncomfortable feeling that the sound is interrupted and the sound quality is deteriorated.

In addition, since a noise signal of a predetermined level is inserted regardless of the voice level of the input side at the time of noise insertion, the voice side section on the playback side may differ due to the difference between the ambient noise of the voice processing unit and the insertion noise. There is a problem in that the sound quality is deteriorated by causing a feeling of strangeness in the sound between the silent section.

Therefore, an object of the present invention is to provide a speech coding apparatus for reducing the discomfort of reproduced sound which occurs when a sound section and a silent section are switched.

[0008]

[Problems to be Solved by the Invention] The above-mentioned problems are shown in FIGS.
This is solved by the configuration of the device shown in. The transmission side has a silence detector 1 that detects a silence section in an input voice signal and outputs silence information, and an encoder 2 that encodes and outputs the silence information and the input voice signal. On the receiving side, the decoding unit 3 which decodes the received silence information and the input voice signal to output the reproduced voice signal and the silence information, and the reproduced voice based on the silence information output from the decoding unit. In a speech coding apparatus having a noise insertion unit 4 ′ that inserts and outputs a predetermined level of noise in a silent section of a signal, in the case of FIG. 1 (Claim 1), 5 is a silence detection unit 1 on the transmission side and a code. The time delay unit is provided between the conversion units 2 and outputs the silence information after delaying a predetermined time from the time when the silence detection unit determines that there is no sound.

Then, the noise inserting section 4 on the receiving side inserts noise of a predetermined level into the silent section of the reproduced voice signal and a certain section immediately before the silent section based on the silence information output from the decoding section, and outputs the noise. To configure.

In the case of FIG. 2 (Claim 2), 5 is provided between the silence detector 1 and the encoder 2 on the transmitting side, and the silence information is delayed by a predetermined time from the time when the silence detector determines that there is no silence. Is a time delay unit that outputs

A mixer 6 is provided on the receiving side and mixes the reproduced voice signal output from the decoding section with a predetermined level of noise in a certain section immediately before the silent section. In the case of FIG. 3 (Claim 3), 5 is provided between the silence detector 1 and the encoder 2 on the transmission side, and outputs the silence information after a predetermined time delay from the time when the silence detector determines that there is no sound. It is a time delay unit.

Reference numeral 7 is provided between the decoding unit 3 and the noise insertion unit 4 on the receiving side, calculates the level of the reproduced voice signal output from the decoding unit, and inserts it from the calculated level of the reproduced voice signal into the silent section. Is a level calculation unit that calculates and outputs a noise level for

Then, the noise insertion unit 4 is configured to insert the noise of the level calculated by the level calculation unit into the silent section of the reproduced voice signal output from the decoding unit and output the noise. In the case of FIG. 4 (Claim 4), 5 is provided between the silence detector 1 and the encoder 2 on the transmission side, and outputs the silence information after a predetermined time delay from the time when the silence detector determines that there is no sound. It is a time delay unit.

Reference numeral 7 is a decoding unit 3 and a noise insertion unit 4 on the receiving side.
The level calculation unit is connected between the two, calculates the level of the reproduced voice signal output from the decoding unit, and calculates and outputs the noise level for inserting into the silent section from the calculated level of the reproduced voice signal.

Numeral 6 is a mixer for mixing the reproduced voice signal output from the decoding section and the noise of the level calculated by the level calculation section in a fixed section immediately before the silent section. The above 6 and 7 are provided in the receiver.

[0016]

In FIG. 1, the time delay unit 5 delays the time when the silence processing is actually applied from the time when the silence detecting unit 1 on the transmitting side determines that there is no sound by a predetermined time.
Therefore, it is possible to avoid performing the silence processing immediately after the audio signal level becomes lower than the predetermined threshold value, and the noise inserting unit 4 on the receiving side can suppress the silence interval of the reproduced audio signal and a constant interval immediately before the silent interval. Since a predetermined level of noise is inserted and output in a section, it is possible to reduce a feeling of strangeness in reproduced sound that occurs when a sound section and a silent section are switched.

In FIG. 2, the mixer 6 mixes a predetermined level of noise to be inserted into the silent section with the reproduced voice signal only in the voiced section including the reproduced voice signal immediately before being judged to be a silent section by the decoder side. . As a result, since the voiced section and the silent section are gradually shifted by overlapping each other, it is possible to reduce the discomfort of the reproduced sound generated when the voiced section and the silent section are switched.

In FIG. 3, the level calculator 8 on the decoder side
Then, the noise level to be inserted into the silent section is calculated from the level of the reproduced voice signal in the sound section immediately before being determined as the silent section. The voiced section immediately before being determined to be a voiceless section is a section that becomes voiced because the time delay unit 5 delays the silence, and is a section to which the silence processing should be originally applied.

Therefore, the reproduced sound in this section is determined to be ambient noise. The average level of this section is calculated, and noise corresponding to that is inserted into the silent section. As a result, the ambient noise in the voiced part and the insertion noise in the silent part are approximated,
It is possible to reduce the discomfort of the reproduced sound that occurs when the sound section and the silent section are switched.

The fourth invention shown in FIG. 4 is the same as that of FIG.
In the method of calculating the insertion noise level from the immediately preceding voiced section described in the operation of (Third invention), the mixing reproduction method of the voiced section and the silent section described in the operation of FIG. 2 (Second invention) is used. It is something to add.

As a result, it is possible to reduce the discomfort of the reproduced sound which occurs when the sound section and the silent section are switched.

[0022]

DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 is a principle diagram of the first invention and a block diagram showing the construction of an apparatus according to an embodiment. FIG. 2 is a principle diagram of the second invention and is a block diagram showing the configuration of the apparatus of the embodiment.

FIG. 3 is a block diagram showing the principle of the third invention and the structure of the apparatus of the embodiment. FIG. 4 is a principle diagram of the fourth invention and is a block diagram showing the configuration of the apparatus of the embodiment. FIG. 5 is a diagram for explaining the mixing method of the embodiment of the fourth invention.

The same reference numerals denote the same objects throughout the drawings. First, an embodiment of the first invention will be described with reference to FIG. In the silence detector 1 on the encoder side shown in FIG. 6A, when the state in which the voice level of the input voice data is less than a predetermined threshold value continues for a certain time or more, it is determined to be a silent section, and the result is the determination result. The silence information is added to the time delay unit 5. When the time delay unit 5 receives the silence information, the time delay unit 5 delays it by a predetermined time (for example, τ) and then adds it to the encoding unit 2 together with the voice data. The encoding unit 2 encodes the silence information to the silent section delayed by a predetermined time (τ), and determines a predetermined voiced section including a section delayed by the predetermined time (τ). The audio data is compression-encoded by the compression encoding method and sent to the transmission path.

On the other hand, in the voice encoding device facing each other through the transmission line, the decoding unit 3 shown in FIG. 1B reproduces the voice data from the compression encoded data and extracts the silence information. , To the noise insertion unit 4. In the noise insertion unit 4,
A noise signal of a predetermined level is inserted and output in the silent section of the reproduced voice data.

As a result, it is possible to avoid the silence processing immediately after the voice level becomes lower than the predetermined threshold value, and the noise insertion unit 4 makes the noise insertion section 4 a silent section and a fixed section immediately before the silent section. Since a predetermined level of noise is inserted and output in, it is possible to reduce the uncomfortable feeling of the voice that occurs at the time of switching between the voiced section and the silent section.

Next, an embodiment of the second invention will be described with reference to FIG. In FIG. 2, the configuration on the encoder side shown in (a) is the same as in the case of the first invention described above, and therefore its explanation is omitted.

In the decoding unit 3 shown in FIG. 2B, the voice data is reproduced from the data compression-coded in the same manner as described above, and the silence information is extracted. The reproduced voice data and the silence information are added to the noise insertion unit 4 and the newly provided mixer 6. Then, the mixer 6 mixes the insertion noise to be inserted into the silent section into the reproduced voice data of the sound section immediately before being determined as the silent section (which may not necessarily match the section of τ described above), It is output only in the voiced section immediately before this silent section.

Further, when there is a silent section or a sound section other than immediately before the silent section, the noise inserting section 4 inserts a predetermined noise signal into the silent section of the reproduced voice data and outputs it. As a result, since the voiced section and the silent section are gradually shifted by overlapping each other, it is possible to reduce the discomfort of the reproduced sound generated when the voiced section and the silent section are switched.

Next, an embodiment of the third invention will be described with reference to FIG. In FIG. 3, the configuration on the encoder side shown in (a) is the same as in the case of the above-described first invention, and therefore its explanation is omitted.

In the decoding section 3 shown in FIG. 3B, the voice data is reproduced from the data compression-coded as described above, and the silence information is extracted. Then, the reproduced voice data is input to the level calculation unit 7, and the level of the reproduced voice data of the voiced section (which does not necessarily coincide with the section of τ described above) immediately before being determined as the silent section. Then, the insertion noise level to be inserted in the silent section is calculated. The calculation result of this insertion noise level is added to the noise insertion unit 4 together with the reproduced voice data. Then, the noise insertion unit 4 generates insertion noise based on the noise level obtained by the above calculation, inserts the insertion noise in the silent section, and outputs it.

The voiced section immediately before being determined to be a voiceless section on the decoder side is the voiced section because the time delay unit 5 delays the silence (in some cases, it does not exactly match). This is the section where the silence processing should be applied. Therefore, the reproduced sound in this section is determined to be ambient noise. The average level of this section is calculated, and noise corresponding to that is inserted into the silent section. As a result, since the ambient noise in the voiced portion and the insertion noise in the voiceless portion are approximated, it is possible to reduce the discomfort of the reproduced sound that occurs when the voiced section and the silence section are switched.

Next, an embodiment of the fourth invention will be described with reference to FIG. In FIG. 4, the configuration on the encoder side shown in (a) is the same as in the case of the above-described first invention, and therefore its explanation is omitted.

In the decoding section 3 shown in FIG. 3B, the voice data is reproduced from the compression-coded data as described above, and the silence information is extracted. Then, the reproduced voice data is temporarily stored in the RAM (not shown) included in the decoding unit 3, and the processing is waited until the next section of the silence information. When it is determined that the next reproduced voice section is silent, the temporarily stored reproduced voice data of the previous section is input to the level calculation unit 7, the average level of the voiced section of the previous section is calculated, and the next silent section is calculated. The insertion noise level of The calculation result of this insertion noise level is added to the noise insertion unit 4 and the mixer 6.

In the fixed section immediately before the silent section, the mixer 6
Insertion noise is generated based on this noise level, and mixed with the reproduced voice data of the voiced section immediately before the silent section added from the decoding unit 3 to obtain output voice data. Next, in the silent section, the noise insertion unit 4 generates insertion noise based on the noise level calculated by the level calculation unit 7, inserts it into the silent section of the reproduced voice data added from the decoding unit 3, and outputs it. . When the next reproduced voice section is voiced, the reproduced voice data added from the decoding unit 3 to the noise inserting unit 4 becomes the output voice data as it is.

The above-mentioned mixing method is carried out by adding (weighting) a window as shown in FIG. 5 to each of the reproduced sound and the insertion noise in the voiced section.
As a result, in the voiced section immediately before the silence section, the reproduced voice signal and the insertion noise are weighted and mixed, respectively, and in the silence section, the insertion noise is inserted as it is, so that the voiced section and the silence section are smoothly connected. It is possible to reduce the discomfort of the reproduced sound that occurs when switching between the sound section and the silent section.

[0037]

As described above, according to the present invention, it is possible to reduce an uncomfortable feeling of reproduced sound which occurs when a sound section and a silent section are switched, and to contribute to the improvement of the sound quality in a speech encoding apparatus. There is a lot to do.

[Brief description of drawings]

FIG. 1 is a principle diagram of the first invention, and a block diagram showing a configuration of an apparatus according to an embodiment;

FIG. 2 is a principle diagram of the second invention, and is a block diagram showing the configuration of the device of the embodiment.

FIG. 3 is a principle diagram of a third invention, and a block diagram showing a configuration of a device according to an embodiment;

FIG. 4 is a principle diagram of a fourth invention, and is a block diagram showing a configuration of a device according to an embodiment;

FIG. 5 is a diagram for explaining a mixing method according to an embodiment of the fourth invention,

FIG. 6 is a block diagram showing a configuration of a conventional speech encoding apparatus.

[Explanation of symbols]

1 is a silence detector, 2 is an encoder, 3 is a decoder, 4,
Reference numeral 4'denotes a noise insertion unit, 5 a time delay unit, 6 a mixer, and 7 a level calculation unit.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Koji Okazaki, 1015 Kamiodanaka, Nakahara-ku, Kawasaki-shi, Kanagawa, Fujitsu Limited (72) Inventor Naoji Matsuo, 1015, Kamikodanaka, Nakahara-ku, Kawasaki, Kanagawa (within Fujitsu Limited) 72) Inventor Toshiaki Nobumoto 3-22-8 Hakataekimae, Hakata-ku, Fukuoka City, Fukuoka Prefecture Fujitsu Kyushu Digital Technology Co., Ltd.

Claims (4)

[Claims] 1. A silence detector (1) for detecting a silence section in an input voice signal and outputting silence information to a transmitting side.
And a coding unit (2) for coding and outputting the silence information and the input voice signal, and the receiving side decodes the received silence information and the input voice signal and reproduces the voice. A decoding unit (3) that outputs a signal and silence information, and noise insertion that inserts and outputs a predetermined level of noise in the silence section of the reproduced audio signal based on the silence information output from the decoding unit. A speech coding apparatus having a section (4 ′), connected between the silence detecting section (1) and the coding section (2) on the transmitting side for a predetermined time from the time when the silence detecting section determines that there is no sound. A time delay unit (5) for delaying and outputting the silence information is provided, and the noise insertion unit (4) on the receiving side, based on the silence information output from the decoding unit, silences the reproduced audio signal. A speech coding apparatus characterized in that a predetermined level of noise is inserted and output in a section and a fixed section immediately before the silent section. Place 2. A silence detecting unit (1) for detecting a silence section in an input voice signal and outputting silence information on the transmitting side.
And a coding unit (2) for coding and outputting the silence information and the input voice signal, and the receiving side decodes the received silence information and the input voice signal and reproduces the reproduced voice. A decoding unit (3) that outputs a signal and silence information, and a noise insertion that outputs by inserting a predetermined level of noise into a silence section of the reproduced voice signal based on the silence information output from the decoding unit A speech coding apparatus having a section (4 ′), connected between the silence detecting section (1) and the coding section (2) on the transmitting side for a predetermined time from the time when the silence detecting section determines that there is no sound. A time delay unit (5) for delaying and outputting the silence information is provided, and the reproduced voice signal output from the decoding unit and noise of a predetermined level are provided to the reception side in a certain section immediately before the silence section. A mixer (6) for mixing is provided, and is output from the mixer in a certain section immediately before the silent section,
Further, the speech coding apparatus is characterized in that the noise insertion section (4) outputs the voiced section except the silent section and a certain section immediately before the silent section. 3. A silence detector (1) for detecting a silence section in an input voice signal and outputting silence information on the transmitting side.
And a coding unit (2) for coding and outputting the silence information and the input voice signal, and the receiving side decodes the received silence information and the input voice signal and reproduces the reproduced voice. A decoding unit (3) that outputs a signal and silence information, and a noise insertion that outputs by inserting a predetermined level of noise into a silence section of the reproduced voice signal based on the silence information output from the decoding unit A speech coding apparatus having a section (4 ′), connected between the silence detecting section (1) and the coding section (2) on the transmitting side for a predetermined time from the time when the silence detecting section determines that there is no sound. A time delay unit (5) for delaying and outputting the silence information is provided, and is connected between the decoding unit (3) and the noise insertion unit (4) on the receiving side, and the reproduced voice output from the decoding unit is provided. A level calculator that calculates the signal level, calculates the noise level to be inserted into the silent section from the calculated level of the reproduced audio signal, and outputs the noise level. Part
(7) is provided, and the noise insertion section (4) inserts and outputs the noise of the level calculated by the level calculation section in the silent section of the reproduced voice signal output from the decoding section. A speech coding apparatus characterized by the above. 4. A silence detecting unit (1) for detecting a silence section in an input voice signal and outputting silence information on the transmitting side.
And a coding unit (2) for coding and outputting the silence information and the input voice signal, and the receiving side decodes the received silence information and the input voice signal and reproduces the reproduced voice. A decoding unit (3) that outputs a signal and silence information, and a noise insertion that outputs by inserting a predetermined level of noise into a silence section of the reproduced voice signal based on the silence information output from the decoding unit A speech coding apparatus having a section (4 ′), connected between the silence detecting section (1) and the coding section (2) on the transmitting side for a predetermined time from the time when the silence detecting section determines that there is no sound. A time delay unit (5) for delaying and outputting the silence information is provided, and is connected between the decoding unit (3) and the noise insertion unit (4) on the receiving side, and the reproduced voice output from the decoding unit is provided. A level calculator that calculates the signal level, calculates the noise level to be inserted into the silent section from the calculated level of the reproduced audio signal, and outputs the noise level. Part
(7) and a mixer (6) for mixing the reproduced voice signal output from the decoding unit and the noise of the level calculated by the level calculation unit in a certain section immediately before the silent section, Output from the mixer in a certain section immediately before the silent section,
A speech coding apparatus characterized in that, in the silent section, the noise of the level calculated by the level calculating section is inserted into the silent section of the reproduced speech signal by the noise inserting section 4 and output.