JP2016178584A - Audio processing device, intercom device and intercom system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To implement comfortable telephone call irrespective of an environment at an installation position.SOLUTION: A voice switch unit 142 is capable of selecting a call transmission mode to give priority to a signal input from a microphone 11a over a signal output to a speaker 12a, and a voice signal detector 141 detects a voice section containing a voice signal from the signal input from the microphone 11a. A voice switch unit 142 selects the call transmission mode when the voice section has been detected in the voice signal detector 141. The voice signal detector 141 adjusts the sensitivity of determining the voice section according to a reverberation time.SELECTED DRAWING: Figure 2

Description

  The present invention relates to a sound processing device, an intercom device, and an intercom system that process a signal input from a microphone.

  Intercoms are often installed in the lobby of apartment buildings. The visitor can tell the visitor to the residents in the dwelling through the intercom installed in the lobby, and the visitor can enter when the resident unlocks. The lobby has a large space and the effect of reverberation increases.

  In the intercom, a voice switch function that prioritizes the larger voice by comparing the voices of the transmitted and received voices may be used (for example, see Patent Document 1).

JP 2012-323865 A

  In a place such as a lobby where the influence of reverberation is large, it becomes difficult to distinguish between voice and noise. Therefore, when a visitor is speaking, there is a case where the voice is erroneously determined as noise, and in this case, the mode that prioritizes transmission is switched from the intermediate mode to the comfortable mode, which may hinder a comfortable call. It was.

  The present invention has been made in view of such circumstances, and an object of the present invention is to provide a voice processing device, an intercom device, and an intercom system that realize a comfortable call regardless of the environment of the installation position.

  In order to solve the above problems, a speech processing apparatus according to an aspect of the present invention includes a voice switch unit capable of selecting a transmission mode in which a signal input from a microphone is given priority over a signal output to a speaker, and an input from the microphone. A voice signal detection unit that detects a voice section including a voice signal from the received signal. The voice switch unit selects the transmission mode when the voice section is detected by the voice signal detection unit, and the voice signal detection unit adjusts sensitivity for determining the voice section according to reverberation time. To do.

  Another aspect of the present invention is an intercom device. This device includes the above-described sound processing device, a microphone that converts collected sound into an electric signal and inputs the electric signal, a speaker that converts an electric signal output from the sound processing device into sound, Is provided.

  Yet another embodiment of the present invention is an intercom system. This intercom system includes an interphone indoor device and an interphone outdoor device for communicating with the interphone indoor device, and an interphone outdoor device in which the above-described interphone device is used.

  It should be noted that any combination of the above-described constituent elements and a representation of the present invention converted between a method, an apparatus, a system, and the like are also effective as an aspect of the present invention.

  According to the present invention, a comfortable call can be realized regardless of the environment of the installation position.

It is a figure which shows the structure of the intercom system which concerns on embodiment of this invention. It is a figure which shows the structure of a lobby intercom apparatus. It is a figure which shows an example of the signal waveform collected by the microphone. It is a figure which shows an example of the moving average value waveform of the signal collected by the microphone. It is a figure which shows another example of the moving average value waveform of the signal collected by the microphone. 6A to 6C schematically show the transition of the level of the speech signal, the transition of the level of the microphone input signal when the influence of reverberation is small, and the transition of the level of the microphone input signal when the influence of reverberation is large. It is the figure drawn. 7A to 7C schematically show the level transition of the speaker output signal, the level transition of the microphone input signal when the influence of reverberation is small, and the level transition of the microphone input signal when the influence of reverberation is large. It is the figure drawn.

  FIG. 1 is a diagram showing a configuration of an intercom system 1 according to an embodiment of the present invention. In the present embodiment, it is assumed that the intercom system 1 is installed in an apartment. The intercom system 1 includes a lobby intercom device 10, a control device 20, an intercom indoor device 30, and a door intercom device 40.

  The lobby intercom device 10 is an intercom outdoor device installed in the lobby, and a visitor inputs a room number and calls the intercom indoor device 30 installed in the visited dwelling unit. The control device 20 is installed in a common space such as a machine room or an administrator room. The control device 20 relays a signal between the lobby intercom device 10 and the intercom indoor device 30. The door intercom device 40 is installed at the entrance of the dwelling unit. The intercom indoor unit 30 and the door intercom unit 40 are in a relationship between a master unit and a slave unit.

  Although only one dwelling unit is illustrated in FIG. 1 for simplification, there are actually a large number of dwelling units, and the control device 20 relays between the plurality of intercom indoor units 30 and the lobby intercom unit 10. The control device 20 is connected to a distribution board and various sensors, and transmits an alert signal to the intercom indoor device 30 of each dwelling unit when an abnormality such as a fire occurs.

  FIG. 2 is a diagram illustrating a configuration of the lobby intercom apparatus 10. The lobby intercom apparatus 10 includes a microphone 11a, a microphone amplifier 11b, an A / D converter 11c, a speaker 12a, a speaker amplifier 12b, a D / A converter 12c, a processing unit 13, a communication unit 17, and an operation unit 18. The processing unit 13 includes an audio processing unit 14, a reverberation time setting unit 15, and a reverberation time estimation unit 16. The audio processing unit 14 includes an audio signal detection unit 141 and a voice switch unit 142.

  In the functional block of the processing unit 13 in FIG. 2, only functional blocks related to the process of interest in the present embodiment are drawn. The function of the processing unit 13 can be realized by cooperation of hardware resources and software resources, or only by hardware resources. Processors, ROM, RAM, and other LSIs can be used as hardware resources. Firmware, applications, and other programs can be used as software resources.

  The microphone 11a collects a visitor's voice, converts it into an electrical signal, and outputs it to the microphone amplifier 11b. The microphone amplifier 11b amplifies the signal input from the microphone 11a and outputs the amplified signal to the A / D converter 11c. The A / D converter 11 c converts the analog signal input from the microphone amplifier 11 b into a digital signal and outputs the digital signal to the audio processing unit 14.

  The communication unit 17 converts an electrical signal (hereinafter referred to as a sound signal) generated by converting sound collected by the microphone 11 a of the lobby intercom device 10 or the intercom indoor device 30 via the control device 20. Is a communication interface for two-way communication. The sound processing unit 14 processes the sound signal received from the intercom indoor device 30 via the communication unit 17 and outputs the processed sound signal to the D / A converter 12c. If the sound signal is compression-encoded, it is decompressed and decoded. Note that the decompression decoding may be performed by a decoding unit (not shown) in the previous stage of the audio processing unit 14.

  The D / A converter 12c converts the digital signal input from the audio processing unit 14 into an analog signal and outputs the analog signal to the speaker amplifier 12b. The speaker amplifier 12b amplifies the signal input from the D / A converter 12c and outputs the amplified signal to the speaker 12a. The speaker 12a converts the electrical signal input from the speaker amplifier 12b into physical vibration and outputs it.

  The operation unit 18 is a user interface for receiving user operations. The operation unit 18 includes at least a numeric keypad button and a call button for inputting a room number.

  The voice switch unit 142 can select a transmission mode in which a signal input from the microphone 11a is given priority over a signal output to the speaker 12a, and a reception mode in which a signal output to the speaker 12a is given priority over a signal input from the microphone 11a. . In the transmission mode, the sound signal input via the communication unit 17 is attenuated. The design may be such that the sound is not output from the speaker 12a after being attenuated by 100%, or the sound level output from the speaker 12a is decreased by being attenuated by an attenuation factor of less than 100%. Moreover, you may use together the process which raises the gain of the microphone amplifier 11b.

  On the other hand, in the reception mode, the signal input from the microphone 11a is attenuated. The design may be such that the signal is attenuated 100% and a significant signal is not sent to the intercom indoor device 30, or the signal level is lowered and sent to the intercom indoor device 30. A process for increasing the amplification factor of the speaker amplifier 12b may be used in combination. The voice switch unit 142 can also select an intermediate mode in which the signal input from the microphone 11a and the signal output to the speaker 12a are attenuated by the same attenuation amount.

  Although FIG. 2 assumes an example in which signal attenuation processing is realized by numerical calculation by digital signal processing, an attenuator constituted by analog elements may be used. In that case, the A / D converter 11 c and the D / A converter 12 c are moved between the voice switch unit 142 and the communication unit 17.

  The audio signal detection unit 141 detects an audio section from a signal input from the microphone 11a. In the following description, “sound” refers to a sound uttered by a person, and the sound signal detection unit 141 determines a section including a sound uttered by a person as a sound section, and only an environmental sound that does not include a sound uttered by a person. Are determined to be non-voice segments. The voice switch unit 142 selects the transmission mode when a voice section is detected by the voice signal detection unit 141, and selects the reception mode or the intermediate mode when a non-voice section is detected.

  Specifically, the audio signal detection unit 141 calculates the long-term moving average value N and the short-term moving average value S of the signal absolute value input from the microphone 11a. The voice signal detection unit 141 calculates a ratio S / N between the short-term moving average value S and the long-term moving average value N. If the ratio S / N is equal to or larger than the set ratio α, the voice signal detecting unit 141 determines the voice section. If it is less than α, it is determined as a non-voice segment. That is, when the short-term moving average value S becomes larger than the long-term moving average value N by a set ratio α or more, it is determined that the voice section is entered. Conversely, when the short-term moving average value S becomes less than the set ratio α of the long-term moving average value N, it is determined to return to the non-voice section.

  Usually, since the amplitude of the audio signal is larger than the amplitude of the noise signal, the amplitude including the audio signal has a larger amplitude. Since the amplitude of both voice and noise fluctuates little by little, in the method of simply determining a voice section when the signal level is high, the determination result is frequently switched between a voice section and a non-voice section. Therefore, hysteresis is given to the signal data using an average value for a certain period. When the state where there is no sound is changed to the state where there is sound, the short-term moving average value S rises before the long-term moving average value N. This is because the long-term moving average value N is a value averaged by including more past data, and therefore changes more slowly than the short-term moving average value S.

  The audio signal detection unit 141 adjusts the sensitivity for determining the audio section according to the reverberation time. Specifically, the sensitivity for determining a speech section is increased as the reverberation time is longer. Reverberation is a phenomenon in which sound can be heard even after the sound source stops vibrating. In general, the reverberation time is defined as the time from when the sound source stops oscillating until it decays by 60 dB. The reverberation time has a property that it becomes longer as the space where the sound is played is wider. Therefore, the reverberation time becomes longer in a large lobby. As the reverberation time becomes longer, the influence of the reverberation added to the sound and collected becomes larger.

  FIG. 3 is a diagram illustrating an example of a signal waveform collected by the microphone 11a. In Figure 3, the same voice (hello) signal waveform when collected by the influence is small environment of reverberation, a signal waveform when collected by the high-impact environment of reverberation depicts schematically. It can be seen that the amplitude fluctuation is smaller in the signal waveform collected in an environment where the influence of reverberation is large. If the fluctuation of the amplitude is small, the difference between the short-term moving average value S and the long-term moving average value N acts in a direction that decreases, so that it is easy to misdetect voice as noise.

  Therefore, when the reverberation time is long, the audio signal detection unit 141 increases the sensitivity to be determined as a voice interval so that it is more easily determined as a voice interval than usual. Hereinafter, three methods for increasing the detection sensitivity of voice detection will be described. In the first method, the audio signal detection unit 141 increases the target period for calculating the long-term moving average value N as the reverberation time is longer.

  FIG. 4 is a diagram illustrating an example of a moving average value waveform of a signal collected by the microphone 11a. In a section voice (hello) is present, towards the short-term moving average waveform is greater than the long-term moving average waveform. Since the long-term moving average value includes a lot of data in the period before the voice is present, it is relatively smaller than the short-term moving average value. Comparing the “long-term moving average waveform” with a relatively long target period with the “long-term moving average waveform” with a relatively short target period, the former is a period before the voice becomes more present Will contain more data. Therefore, the former has a lower level than the latter. Therefore, the use of the “long-term moving average value” having a relatively long target period has a larger difference from the short-term moving average value, and is easily determined as a voice section. The longer the reverberation time, the smaller the difference between the long-term moving average value and the short-term moving average value. However, by extending the target period of the long-term moving average value, the influence can be offset or mitigated.

  In the second method of increasing the detection sensitivity of voice detection, the voice signal detection unit 141 shortens the target period for calculating the short-term moving average value S as the reverberation time is longer.

  FIG. 5 is a diagram showing another example of the moving average waveform of the signal collected by the microphone 11a. In FIG. 5, when comparing the “short-term moving average waveform” with a relatively short target period and the “short-term moving average waveform” with a relatively long target period, the former has a larger amplitude fluctuation. Therefore, the use of the “short-term moving average value” having a relatively short target period tends to increase the difference from the long-term moving average value, and it is easy to determine the voice section. The longer the reverberation time, the smaller the difference between the long-term moving average value and the short-term moving average value. However, by reducing the target period of the short-term moving average value, the influence can be offset or mitigated.

  In the third method of increasing the detection sensitivity of the sound detection, the sound signal detection unit 141 decreases the setting ratio α as the reverberation time is longer. If the setting ratio α is reduced, the ratio S / N between the short-term moving average value S and the long-term moving average value N is likely to be greater than or equal to the setting ratio α, and it is easy to determine a voice section.

  In the above description, each of the target period of the short-term moving average value S (that is, the number of data samples), the target period of the long-term moving average value N, and the setting ratio α are values derived from experiments and simulations. Can be used. Moreover, the relationship derived | led-out based on experiment and simulation can also be used for the relationship between the change value of reverberation time, and the change value of the target period of the long-term moving average value N. FIG. The relationship may be defined by a function, or a table describing the relationship may be used. The same applies when the target period of the long-term moving average value N or the set ratio α is changed.

  Further, the relationship between the reverberation time and the target period of the long-term moving average value N may be defined by the mode. For example, three modes of “short”, “normal”, and “long” are prepared for the reverberation time, and three modes of “short”, “normal”, and “long” are prepared for the target period of the long-term moving average value N. . When the reverberation time mode is set to “long”, the audio signal detection unit 141 calculates the long-term moving average value N by the number of samples when the target period is “long”.

  Returning to FIG. The reverberation time setting unit 15 sets the reverberation time in the audio signal detection unit 141. The reverberation time setting unit 15 sets the value of the reverberation time input by the user from the operation unit 18 in the audio signal detection unit 141. For example, at the time of construction of the lobby intercom apparatus 10, the contractor may measure the reverberation time of the lobby using a sound level meter and input the measured value from the operation unit 18.

  Thus, although the value input by the user can be used for the reverberation time, an estimated value can also be used. In addition, using estimated values can immediately reflect changes in environmental conditions such as wall material changes, temperature, and humidity in the reverberation time. The reverberation time estimation unit 16 analyzes the signal input from the microphone 11a and estimates the reverberation time. For example, the reverberation time estimation unit 16 can estimate the reverberation time by detecting the inclination from the end of the voice section of the signal input from the microphone 11a.

  6A to 6C schematically show the transition of the level of the speech signal, the transition of the level of the microphone input signal when the influence of reverberation is small, and the transition of the level of the microphone input signal when the influence of reverberation is large. It is the figure drawn. Comparing FIG. 6 (b) and FIG. 6 (c), the microphone input signal when the influence of reverberation shown in FIG. For example, the reverberation time estimation unit 16 estimates the reverberation time by measuring the time from when the speech section is switched to the non-speech period until the level of the microphone input signal reaches the average level of the non-speech period. In such a method using the speech voice of a visitor, it is not necessary to output sound from a speaker 12a as will be described later, and it is not necessary to prepare test sound source data.

  The microphone input signal used for estimating the reverberation time is not limited to the signal including the speech sound. For example, the reverberation time estimation unit 16 may output sound from the speaker 12a and collect the sound from the microphone 11a to estimate the reverberation time. In this case, since the end timing of the output sound from the speaker 12a can be specified precisely, the reverberation time can be estimated with higher accuracy.

  7A to 7C schematically show the level transition of the speaker output signal, the level transition of the microphone input signal when the influence of reverberation is small, and the level transition of the microphone input signal when the influence of reverberation is large. It is the figure drawn. Comparing FIG. 7B and FIG. 7C, the inclination of the microphone input signal when the influence of the reverberation shown in FIG. When the speaker output signal is used, the reverberation time estimation unit 16 estimates the reverberation time by measuring the time from the end of the speaker output until the level of the microphone input signal reaches the average level of the non-voice period.

  The reverberation time estimation unit 16 may include a test sound source data holding unit 16a that holds sound source data of a reverberation estimation test signal output from the speaker 12a. The reverberation time estimation unit 16 reads sound source data from the test sound source data holding unit 16a and outputs a test sound from the speaker 12a in the activation sequence of the sound processing unit 14. The reverberation time estimation unit 16 collects the test sound with the microphone 11a and estimates the reverberation time. Since the reverberation time is also affected by temperature and humidity, estimating the reverberation time each time the sound processing unit 14 is activated is a desirable control from the viewpoint of using the reverberation time that matches the environment at that time.

  It should be noted that as the sound source data of the reverberation estimation test signal, back tone sound source data that rings when a visitor calls a resident may be used. In this case, it is not necessary to separately prepare sound source data for the reverberation estimation test signal. The reverberation time estimation unit 16 may estimate the reverberation time by collecting a ringing tone from the microphone 11a each time the call button is pressed.

  The reverberation time estimation unit 16 may collect the received voice during a call output from the speaker 12a from the microphone 11a and estimate the reverberation time. This estimation process is preferably executed when the voice switch unit 142 selects the reception mode. Also in this case, it is not necessary to separately prepare sound source data for the reverberation estimation test signal.

  As described above, according to the present embodiment, the switching accuracy between the transmission mode and the reception mode is increased and comfortable by adjusting the sensitivity of the audio signal detection unit 141 according to the reverberation time, which is determined as the audio section. A call can be realized. This is particularly effective for intercom devices installed in places such as lobbies where the effects of reverberation are significant. Further, by providing the reverberation time estimation unit 16, it is possible to reflect the environmental change of the installation location in the voice switch function in real time.

  The present invention has been described based on the embodiments. The embodiments are exemplifications, and it will be understood by those skilled in the art that various modifications can be made to combinations of the respective constituent elements and processing processes, and such modifications are within the scope of the present invention. .

  In the above embodiment, the method of determining a voice section when the ratio S / N between the short-term moving average value S and the long-term moving average value N is greater than or equal to the set ratio α has been described. If the difference (S−N) between the long-term moving average value N and the long-term moving average value N is equal to or larger than the set value β, it may be determined that the voice section.

  In the above embodiment, the example in which the voice processing unit 14 including the voice switch unit 142 is mounted on the lobby intercom apparatus 10 has been described. In this regard, the sound processing unit 14 may be mounted on the door intercom device 40 or the intercom indoor device 30. Note that the sound processing unit 14 for reducing the influence of reverberation collected by the microphone of the door intercom device 40 may be mounted on the intercom indoor device 30 serving as a master unit. Since the corridor where the door intercom device 40 is installed is also a place where the reverberation is relatively large, it is significant to mount the audio processing unit 14.

  Further, when the space in the dwelling unit is wide, such as when the ceiling in the dwelling unit is high, the sound processing unit 14 for reducing the influence of reverberation collected by the microphone of the intercom indoor unit 30 may be mounted on the intercom indoor unit 30. Good.

  Further, the audio processing unit 14 may be mounted on the control device 20. When the control device 20 relays a sound signal between the lobby intercom device 10 and the intercom indoor device 30, a voice switch may be applied. In that case, the reverberation time estimation unit 16 may be mounted on the control device 20 or the lobby intercom device 10. In the latter case, the reverberation time estimated by the reverberation time estimation unit 16 is transmitted to the control device 20 via the trunk line.

  In the above embodiment, the intercom system 1 installed in a condominium has been described. However, the present invention can be similarly applied to an intercom system 1 installed in a detached house. In the intercom system 1 installed in a detached house, the lobby intercom device 10 and the control device 20 are not provided. The voice switch function described above is applied to a sound signal between the intercom indoor device 30 and the door intercom device 40.

  The embodiment may be specified by the following items.

[Item 1]
A voice switch unit 142 capable of selecting a transmission mode in which a signal input from the microphone 11a is prioritized over a signal output to the speaker 12a;
An audio signal detector 141 that detects an audio section including an audio signal from a signal input from the microphone 11a;
The voice switch unit 142 selects the transmission mode when the voice section is detected by the voice signal detection unit 141,
The audio signal detection unit 141 adjusts the sensitivity to determine the audio section according to the reverberation time.
Thereby, it can suppress that the precision which determines with a speech area by reverberation falls.
[Item 2]
The sound processing device (sound processing unit 14) according to item 1, wherein the sound signal detecting unit 141 increases sensitivity for determining a sound section as the set reverberation time is longer.
Thereby, it is possible to reduce the probability of erroneously determining a speech segment as a non-speech segment due to reverberation.
[Item 3]
The voice signal detection unit 141 determines a voice section when a short-term moving average value of a signal input from the microphone 11a is larger than a long-term moving average value by a threshold or more. 3. The speech processing apparatus (speech processing unit 14) according to item 1 or 2, characterized in that the target period for calculating the moving average value is lengthened.
Thereby, a long-term moving average value can be shifted to the lower side, and the sensitivity for determining a voice segment can be increased.
[Item 4]
The voice signal detection unit 141 determines a voice section when a short-term moving average value of a signal input from the microphone 11a is larger than a long-term moving average value by a threshold or more. 3. The speech processing apparatus (speech processing unit 14) according to item 1 or 2, wherein a target period for calculating the moving average value is shortened.
Thereby, the change of a short-term moving average value can be enlarged, and the sensitivity which determines with a speech area can be improved.
[Item 5]
The voice signal detection unit 141 determines a voice section when a short-term moving average value of a signal input from the microphone 11a is greater than or equal to a threshold value than a long-term moving average value, and the threshold value increases as the set reverberation time is longer. The audio processing device (audio processing unit 14) according to item 1 or 2, characterized in that
Thereby, it is possible to increase the probability that the ratio or difference between the short-term moving average value and the long-term moving average value is equal to or higher than the set ratio or the set value, and to increase the sensitivity for determining the voice section.
[Item 6]
6. The voice processing device (voice processing unit 14) according to any one of items 1 to 5, wherein the reverberation time is set by user input.
Thereby, the reverberation time can be set to a value according to the installation environment.
[Item 7]
The speech processing apparatus according to any one of Items 1 to 5, wherein the reverberation time is set by a reverberation time estimation unit that analyzes a signal input from the microphone 11a and estimates the reverberation time. Voice processing unit 14).
Thereby, the reverberation time can be set to a value reflecting the installation environment in real time.
[Item 8]
The voice processing device (voice processing unit 14) according to any one of items 1 to 7,
A microphone 11a that converts the collected sound into an electric signal and inputs the electric signal to the sound processing device (sound processing unit 14);
A speaker 12a for converting an electrical signal output from the audio processing device (audio processing unit 14) into sound;
An intercom device (lobby intercom device 10).
Thereby, it can suppress that the precision which determines with an audio | voice area falls by reverberation, and can implement | achieve the intercom apparatus in which a comfortable telephone call is possible.
[Item 9]
An intercom indoor device (interphone indoor device 30);
An intercom outdoor device (lobby interphone device 10) for communicating with the intercom indoor device (interphone indoor device 30), an interphone outdoor device (lobby interphone device 10) in which the interphone device according to item 8 is used,
An intercom system 1 comprising:
Thereby, it can suppress that the precision which determines with an audio | voice area falls by reverberation, and can implement | achieve the intercom system in which a comfortable telephone call is possible.

  DESCRIPTION OF SYMBOLS 1 Intercom system, 10 Lobby intercom apparatus, 11a microphone, 11b Microphone amplifier, 11c A / D converter, 12a Speaker, 12b Speaker amplifier, 12c D / A converter, 13 Processing part, 14 Sound processing part, 141 Sound signal detection Unit, 142 voice switch unit, 15 reverberation time setting unit, 16 reverberation time estimation unit, 16a test sound source data holding unit, 17 communication unit, 18 operation unit, 20 control device, 30 intercom indoor device, 40 door intercom device.

Claims (9)

A voice switch unit capable of selecting a transmission mode in which a signal input from a microphone is given priority over a signal output to a speaker;
A voice signal detection unit that detects a voice section including a voice signal from a signal input from the microphone;
The voice switch unit selects the transmission mode when the voice section is detected in the voice signal detection unit,
The voice processing apparatus, wherein the voice signal detection unit adjusts sensitivity for determining the voice section according to reverberation time.   The audio processing apparatus according to claim 1, wherein the audio signal detection unit increases sensitivity for determining a voice interval as the set reverberation time is longer.   The voice signal detection unit determines a voice section when a short-term moving average value of a signal input from the microphone is greater than a long-term moving average value by a threshold or more, and the longer the reverberation time is set, the longer the long-term movement is. 3. The speech processing apparatus according to claim 1, wherein a target period for calculating the average value is lengthened.   The voice signal detection unit determines a voice section when a short-term moving average value of a signal input from the microphone is larger than a long-term moving average value by a threshold or more, and the shorter the reverberation time is set, the shorter the short-term movement is. The speech processing apparatus according to claim 1, wherein a target period for calculating the average value is shortened.   The voice signal detection unit determines a voice section when a short-term moving average value of a signal input from the microphone is greater than or equal to a threshold value than a long-term moving average value, and decreases the threshold value as the set reverberation time is longer. The speech processing apparatus according to claim 1 or 2, wherein   The speech processing apparatus according to claim 1, wherein the reverberation time is set by user input.   The speech processing apparatus according to claim 1, wherein the reverberation time is set by a reverberation time estimation unit that analyzes a signal input from the microphone and estimates the reverberation time. A voice processing device;
A microphone that converts the collected sound into an electrical signal and inputs it to the sound processing device;
A speaker that converts an electrical signal output from the audio processing device into sound;
An intercom apparatus comprising: Intercom indoor unit,
An intercom outdoor device for communicating with the intercom indoor device, wherein the intercom outdoor device in which the interphone device according to claim 8 is used;
An intercom system comprising:

Images