JPH04119028A - Voice detector - Google Patents

Abstract

PURPOSE:To always obtain a proper background noise power by receiving a reception packet presence signal from a packet decomposing section and making the result of calculation of an estimate silence time power calculation section for a period when a reception packet is in existence equal to the result of calculation latched at a preceding period. CONSTITUTION:A reception packet presence signal inputted from an input terminal 24 is inputted to a switch 25, which selects an output of a block power calculation section 15 when the reception voice packet is absent and selects an output of an estimate silence time power calculation section 20 when the packet is present and outputs the selected output to a calculation section 20. Thus, the operation of the calculation section 20 is the same as that of a conventional system when no reception packet exists, but the power in the silence state is kept to that calculated for a preceding period regardless of the block power when the packet is in existence. Thus, even when a signal including a complete silence period like an output signal of an echo canceller 5 is inputted, it is possible to always obtain a proper background noise power by the calculation section 20, and the voice detector is realized, in which background noise is not mis-discriminated to be a voiced sound.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a voice detector used in an audio signal packet assembly and disassembly device to determine the presence or absence of an audio signal.

(Prior Art) With the recent diversification of media, there is an increasing demand for multimedia communication that integrates different media information such as voice, data, and images. One method of media integration is multimedia communication using a packet switching method, and along with this, research on packetization of audio signals, which is one type of media information, is progressing.

Figure 6 is the Electronic Information and Communication Handbook (Ohmsha, 1932).
This is an example of the configuration of a voice packet assembly/disassembly device. In the figure, (1) is the subscriber, (2
) is a voice packet assembling and disassembling device, and this voice packet assembling and disassembling device (2) has a hybrid circuit (3) that inputs a call signal from a subscriber (1), and converts the analog signal from this hybrid circuit (3) into PCM. Convert to signal ^/
An echo canceller (5) that removes echo components included in the signal input from the D converter (4) and the ^/D converter (4).
), an encoding unit (6) which obtains the output of the echo canceller (5), performs high-efficiency encoding of the audio signal, and outputs it to the packet assembly unit (8), and a coding unit (6) that detects the presence or absence of the audio signal and outputs the detected signal. a voice detector (7) outputting to the packet assembly unit (8);
Using the detection signal input from the voice detector (7), the packet assembling unit (8) packetizes the encoded voice signal input from the encoder (6) only when there is a voice, and outputs the packet to the packet network. ) and a packet switch (9
), a packet disassembly unit (10) that removes header information from received packets input from the input packet, a silence insertion unit (11) that inserts silence at times when there are no received packets, and absorbs fluctuations in packet transmission delay time in the packet communication network. a fluctuation absorption buffer (12), an echo canceller (5) that decodes the encoded audio signal into a PCM signal, and a D/
A decoding unit (13) outputting to the A converter (14), the PCM
It is equipped with a D/A converter (14) that converts the signal into an analog signal and outputs it to the subscriber <1) via the hybrid circuit (3).
) is configured to utilize the line effectively through high-efficiency speech encoding and silence compression.

Here, in order to perform silence compression, a voice detector (7) is required, and conventionally, for example, a voice detector as shown in FIG. 7 has been proposed. This is Japanese Patent Publication No. 60-1178
This is disclosed in Publication No. 38, and receives a sampled and quantized input signal, and sends out a determination output that determines whether the input signal is audible or silent, and the determination output is, for example, It changes every approximately 6 ms, the minimum time interval that can be determined (hereinafter referred to as an interval).

That is, the voice detector (7) having the illustrated configuration includes a section power calculation section (15) that adds the power of the input signal within the section;
A zero-crossing counting unit (16
), 1-section delay section (17) that holds the power of 1 section before.
, the output of the section power calculation section (15) and the one section delay section (3
) is about twice or more, it is assumed that there is a sound signal and the output is set to 1.The output of the judgment unit (18) based on the ratio to the previous power and the section power calculation unit (15) is absolute. Judgment unit (19
), which updates the value of silence power for each section based on the output of the section power calculation section (15) when the utterance/silence determination section (9), which will be described later, is in the silence determination state or the hangover state.
Estimated silent power calculation unit (20) that increases the silent power to the slope of the normal voice power level when the interval power is larger than the silent power, and replaces the silent power with the interval power when it is smaller; estimated silent power calculation unit Part (20
) and the pressure of the section power calculation section (15) is about 3 times or more, the pressure is set to 1. The determination section (21) based on the ratio of the silent power and the section power, the τ crossover counting section ( 16
) Judgment unit (22) based on zero crossing which sets the pressure to 1 if there is a number of pressures that seem to be silent, a judgment unit (18) based on the ratio to the previous power, and judgment based on the ratio of silence power to section power. (21), a determination unit (19) based on the ratio of the absolute value level and the section power, and a sound determination unit (23) that outputs a determination result of utterance or silence based on the pressure results of the determination unit (22) based on zero crossing. ).

Next, let us consider the operation when the conventional voice detector (7) as described above is used in the voice packet assembly/disassembly device (2) shown in FIG. In FIG. 6, the voice detector (7
) is output from the echo canceller (5). The function of the echo canceller (5) is that of the packet disassembly unit (
10), silence insertion part (11), fluctuation absorption buffer (1
2) The received audio signal output from the D/A converter (14) via the decoding section (13) is used to cancel echo signals generated when the received audio signal goes around the hybrid circuit (3).

Fig. 8 is a diagram for explaining the operation of the echo canceller (5), and output signals when Fig. 8 (a) is used as the near-end input signal of the echo canceller are shown in Fig. 8 (b) and (C). It is. Normally, the echo canceller (5) uses NLP (Non-
FIG. 8(b) shows the output signal when the NLP function is not activated, and FIG. 8(C) shows the output signal when the NLP function is activated. Figure 8 f
If a) is the near-end input signal of the echo canceller (5) in FIG. 6, the low-level signals in ranges (I) and (III) are background noise signals from the subscriber (1), The high-level signal in range (II) is a signal in which a background noise signal from the subscriber (1) and an echo signal generated when the received voice signal loops around in the hybrid circuit (3) are superimposed. Although the call voice signal from subscriber (1) is not included here, it is omitted because it is not particularly necessary for future explanations, and ranges (I) and (11).

In (III), subscriber (1) is assumed to be in a silent state. Such a signal is an echo canceller (5)
As shown in Figure 8(b), when the NLP function is not activated, the echo signal is suppressed in range (II) and its level becomes low, as shown in Figure 8(C). When the NLP function is activated as shown in FIG. 1, the residual echo that cannot be completely eliminated in range (II) is completely canceled out, resulting in complete silence.

However, the conventional sound pressure detector (7) shown in FIG.
As an input, an echo canceller (
When the output signal of 5) is given, if the length of the range (H) is sufficiently long compared to the section that is the operating time unit of the sound detector (7), there is complete silence in the range (II), so The interval power calculation unit (15) calculates τ, and the estimated silent power calculation unit (20) also calculates zero as the silent power. Next, when the range (III) is entered and background noise begins to be input to the speech detector (7), the output of the section power calculation section (15) calculates and outputs the power of the background noise. As a result, the ratio between the output of the estimated silent power calculation section (20) and the output of the section power calculation section (15) increases, and the determination section (21) based on the ratio of the silent power to the section power increases.
), the output is set to 1, and even though the input signal is background noise, it will be erroneously determined to be a voice.

(Problem to be Solved by the Invention) As described above, the conventional audio pressure detector (a) is capable of estimating when a signal including a complete silent section, such as the pressure signal of the echo canceller (5), is given as input. The silent power calculation unit (20) calculates zero in a completely silent section, and this is different from the original background noise power, so even though the signal following the complete silent section is background noise, it is There was a problem where it could be incorrectly determined that there was a sound.

This invention was made to solve the above-mentioned problems, and even when a signal including a complete silent section, such as the output signal of an echo canceller, is input, the background noise can be suppressed. The purpose of this invention is to obtain a voice detector that does not make such a judgment.

[Means to solve the problem]

The voice detector according to the present invention digitally converts an input voice signal from a subscriber through a hybrid circuit, and then
The echo canceller removes the echo component contained in the input signal, packetizes the audio signal only when there is sound based on the audio presence/absence detection signal, and outputs the packet to the packet network, and also receives input from the packet network. The header information is removed from the packet by the packet disassembly unit, silence is inserted during the time when there is no received packet, and after absorbing fluctuations in packet transmission delay time, it is converted to analog and sent to the subscriber via the above hybrid circuit. The audio presence/absence detection means used in the packet assembly/disassembly device that outputs the signal includes a first power calculation means for calculating the power of the input signal from the echo cancellation, and an output of the first power calculation means. a second power calculation means for calculating the background noise power during silence based on the background noise power, and comparing the power calculated by the first power calculation means and the power calculated by the second power calculation means, A voice detector is provided between the first and second power calculation means, and inputs received packet presence/absence information from the packet disassembly section, and determines the presence or absence of a received packet. The second power calculating means is provided with a control means for controlling the second power calculating means to output the same value as the background noise power during silence calculated in the past.

[Effect]

In this invention, by inputting the presence/absence signal of a received packet from the packet disassembly section, the control means maintains the calculation result of the estimated silent power calculation section in the previous section in the section where there is a received packet. The result is the same as the calculated result. In other words, the range (II
), although there is an echo signal generated by the received audio signal looping around in the hybrid circuit, the received packet presence/absence signal output from the bucket)-decomposition unit is in the presence state, and therefore, the reception By making the calculation result of the estimated silent power calculation unit in the interval with a packet the same as the calculation result held in the previous interval, the estimated silence can be calculated even in the interval where the output signal of the echo canceller is the same as the calculation result held in the previous interval. It becomes possible to hold an appropriate background noise power in the time power calculation section.

〔Example〕

An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 shows a voice packet assembly and disassembly device for explaining the present invention in detail, and in the figure, (1) to (6) (8)
.about.(14) are the same as those shown in FIG. 6, but the voice detector (7) is configured to input the received packet presence/absence signal from the packet disassembly section (10).

Further, FIG. 2 is a block diagram of a voice detector showing an embodiment of the present invention, and (15) to (23) are the same as the conventional example shown in FIG. (24) is the packet decomposition part (
an input terminal for inputting a received packet presence/absence signal from lO);
(25) is a control means that switches the contact depending on the presence or absence of a received packet, and controls the second power calculation means to output the same value as the background noise power during silence calculated in the past during the time when there is a received packet. It is a switch that performs

Further, FIG. 1 shows an example in which the embodiment of the present invention shown in FIG. 2 is used in an audio packet assembly and disassembly device. In the figure, (1) to (6) and (8) to (14) are the same as those shown in FIG. It is configured to input.

Next, the operation of the specific embodiment shown in FIG. 2 will be explained. (15) to (19) and (21) to (23) are the same as in the conventional example, and therefore will be omitted. Input terminal (24)
The received packet presence/absence signal input from the switch (2
5), the switch (25) selects the pressure of the interval power calculation unit (15) when there is no received voice packet, and selects the output of the estimated silent power calculation unit (20) when there is a received voice packet, and selects the output of the estimated silent power calculation unit (20) to calculate the estimated silent time. Output to the power calculation section (2o). Therefore, the operation of the estimated silent power calculation unit (2o) is the same as the conventional example when there is no received packet, but when there is a received packet, the silent power is calculated in the previous interval regardless of the interval power. The power is retained when there is no sound.

Therefore, even if a signal including a complete silent period is input, such as the output signal of the echo canceller (5), the estimated silent power calculation unit (2o) can always calculate an appropriate background noise power, A voice detector that does not erroneously judge noise as voice presence can be obtained.

Further, the present invention provides a voice detection system that includes means (21) for determining the presence or absence of voice based on the ratio of the output of the estimated silent power calculation unit (2o) and the section power, as in the conventional example shown in FIG. Estimated silent power calculation unit (20)
The present invention is also effective for a voice detector equipped with means for generating a threshold value based on the output of , and determining the presence or absence of voice based on this threshold value and the magnitude of the section power.

For example, consider a voice pressure detector that includes a determining section configured as shown in FIG. 3 instead of the determining section (21) based on the ratio of silent power to section power shown in FIG. In the figure, (211) is a threshold generation unit that generates a threshold based on the output of the estimated silent power calculation unit, and (212) is a determination unit that determines the magnitude of the threshold and the section power. A section power calculation unit (15) and an estimated silent power calculation unit (20) related to the operation are also shown. The threshold generation section (211) is the estimated silent power calculation section (20).
According to the output of
The output is set to 1 when the section power is greater than the output threshold in 1). %8 as input signal Figure (c
) Given the echo canceling output signal shown in
As in the conventional example shown in FIG.
When background noise starts to be input again in II), the section power becomes stronger than the pressure of the threshold generation section (211,), and the judgment section (212) based on the magnitude of the threshold and section power becomes 1. Output. For this reason, as with the conventional voice detector shown in FIG. 7, background noise is erroneously determined to be voiced.

FIG. 4 shows an embodiment in which the present invention is applied to solve this problem. In the figure, the switch (25) operates similarly to the switch (25) of the embodiment of the present invention shown in FIG. Therefore, it can be seen that the estimated silent power calculation unit (20) can always calculate the background noise appropriately, and there is no possibility of erroneously determining the presence of speech as background noise.

Furthermore, although FIG. 1 shows an example in which the voice detector according to the present invention is applied to a voice packet assembly and disassembly device, as shown in FIG. It is also applicable to voice packet assembly and disassembly devices that do not use.

(Effects of the Invention) As described above, according to the present invention, the received packet presence/absence signal is input from the packet disassembly section, and in the section where there is a received packet, the calculation result of the estimated silent power calculation section is calculated from the previous section. Since the calculation result is the same as that held in the Estimated Silence Power Calculator, even if a signal including a complete silent section is input, such as an echo canceller output signal, the estimated silent power calculation section will always calculate an appropriate background noise power. This has the effect of providing a voice detector that does not erroneously determine background noise as a voice.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of a voice detector according to the present invention applied to a voice packet assembly and disassembly device, FIG. 2 is a block diagram of a voice detector according to an embodiment of the present invention, and FIG. 3 is an estimated silence power FIG. 4 is a block diagram of a determination unit that determines utterance and non-utterance based on the magnitude of the threshold value generated based on the output of the calculation unit and the interval power, and its related parts. FIG. 5 is a block diagram of a second embodiment that has been used, and FIG. 5 is a block diagram of an example in which the voice detector according to the present invention is applied to a voice packet assembly and disassembly device that does not use a high-efficiency voice encoder and decoder. 7 is a block diagram of a voice packet assembly and disassembly device, FIG. 7 is a block diagram of a conventional voice detector, and FIG. 8 is an explanatory diagram of an example of an input pressure signal of an echo canceller. In the figure, (1) is a subscriber, (2) is a voice packet assembly/disassembly device, (3) is a hybrid circuit, and (4) is an A/
D converter, (5) is an echo canceller, (A) is a voice detector, (8) is a packet assembly unit, (9) is a packet switch, (10) is a packet disassembly unit, (11) is a silence insertion unit, (12) is a fluctuation absorption buffer, (14) is a D/A converter, (15) is an interval power calculation unit, (20) is an estimated silent power calculation unit, and (21) is a combination of silent power and interval power. (24) is a reception packet presence/absence signal input terminal, (25) is a switch, (211) is a threshold value generation section, and (212) is a determination section based on the magnitude of the threshold and section power. In the drawings, the same reference numerals indicate the same or corresponding parts.

Claims (1)

[Claims] After converting the input audio signal from the subscriber into a digital signal through a hybrid circuit, an echo canceler removes the echo component contained in the input signal, and generates an audio signal only when there is sound based on the audio presence/absence detection signal. In addition to packetizing and outputting to the packet network, the packet disassembly section removes header information from the received packets input from the packet network, inserts silence during times when there are no received packets, and suppresses fluctuations in packet transmission delay time. After absorption, the power of the input signal from the echo cancellation is calculated as means for detecting the presence or absence of the audio used in a packet assembly/disassembly device that converts the audio signal into analog and outputs the audio signal to the subscriber via the hybrid circuit. a first power calculation means; a second power calculation means for calculating background noise power during silence based on the output of the first power calculation means; and a power calculated by the first power calculation means. and a determination means for determining the presence or absence of a voice by comparing the power calculated by the second power calculation means, the voice detector provided between the first and second power calculation means. Then, information on the presence or absence of a received packet is input from the packet decomposition unit, and the second power calculation means is controlled to output the same value as the previously calculated background noise power during silent periods during the time when there is a received packet. A voice detector characterized in that it is equipped with a control means.