JP4914319B2 - COMMUNICATION VOICE PROCESSING METHOD, DEVICE THEREOF, AND PROGRAM THEREOF - Google Patents

Description

  The present invention relates to an audio processing method, an apparatus thereof, and a program thereof in video communication and audio communication that connect remote locations by communication.

  In recent years, with the development of communication networks, a communication environment for connecting remote locations with always-connected lines is being developed. These always-on environments are used as watching services at home or for family communication between remote locations. In the workplace, it is used when performing telework by working together from remote locations or working from home. However, in a constantly connected environment, both communication situations are always picked up by a microphone, and information is collected by various sensors. This causes a problem that both the sender and the listener infringe on privacy.

Conventionally, a video communication apparatus disclosed in Patent Document 1 is known as a method for performing always-connected communication while avoiding this problem in an always-connected environment. A conventional video communication apparatus will be briefly described with reference to FIG. The video communication apparatus 100 is a communication terminal, and includes input means 10, output means 20, communication control means 30, terminal status information processing means 40, input / output control processing means 50, audio processing means 60, video. And processing means 70. The input unit 10 includes an imaging unit 11, a voice input unit 12, a proximity detection unit 13, and an operation unit 14. The proximity unit 13 is a distance sensor using infrared rays, ultrasonic waves, or the like, and measures the distance between the video communication apparatus 100 and the user. The proximity information detected by the proximity unit 13 is input to the terminal status information processing unit 40. The terminal status information processing unit 40 controls the operations of the video processing unit 70 and the audio processing unit 60 via the input / output control processing unit 50 based on the proximity information. The video processing means 70 performs processing for reducing the sharpness of the video according to the control signal from the input / output control processing means 50. Similarly, the sound processing means 60 performs a process for reducing the definition of the sound or a process for reducing the volume. When the distance between the user and the video communication device 100 is short and the user wants to actively communicate with the communication partner, the clarity of video and audio is increased. Conversely, when the distance between the user and the video communication device 100 is long and reluctant to communicate, the definition of video and audio is reduced or the volume is reduced.
JP 2006-140747 A (FIG. 1)

  There is no description in Patent Document 1 regarding a specific process for reducing the definition of the voice of a conventional communication device. In general, when the sharpness of the voice is lowered, it becomes difficult to hear, the listener often feels uncomfortable, and the usability as a communication device is poor. Further, in the case of a method of reducing the voice, if the listener listens and listens, the content of the conversation can be understood. As described above, the conventional communication apparatus is insufficient from the viewpoint of privacy protection and usability.

  The present invention has been made in view of these points. In particular, an object is to provide a communication speech processing apparatus, a method thereof, and a program thereof that can focus on the speech processing apparatus and can grasp the other party's situation while avoiding privacy infringement and that do not make the listener uncomfortable.

The communication speech processing apparatus according to the present invention includes a frequency conversion unit, a clarity setting unit, a band division unit, a feature amount calculation unit, a filter unit, and a frequency inverse conversion unit. The frequency converter converts the input audio signal into a frequency domain signal. The articulation setting unit sets articulation. The band dividing unit performs frequency analysis of the input audio signal and generates a band divided signal divided into predetermined frequency bands. The feature amount calculation unit calculates a feature amount for each band division signal. The filter unit multiplies a predetermined acoustic frequency signal other than the feature amount of the band division signal and the input audio signal for each of a plurality of frequency bands . The frequency inverse transform unit transforms the output signal of the filter unit into a time domain signal.

  The communication voice processing apparatus according to the present invention outputs as an output signal an acoustic signal such as noise, environmental sound, and musical sound filtered by the feature amount of the input voice signal collected by the microphone unit. Therefore, by making the sound signal a comfortable sound, the privacy of the sender (self) can be protected without making the listener uncomfortable. Further, although the details of the other party's voice are concealed by the acoustic signal filtered by the feature quantity of the input voice signal, it is possible to convey the atmosphere of the situation including the conversation while protecting the privacy. Furthermore, since the communication sound processing apparatus of the present invention performs processing using only the input signal of the microphone, it can be easily incorporated into a communication apparatus or system.

  Embodiments of the present invention will be described below with reference to the drawings. The same reference numerals are given to the same components in a plurality of drawings, and the description will not be repeated.

FIG. 1 shows an example of a functional configuration of the communication voice processing apparatus according to the first embodiment of the present invention. The operation flow is shown in FIG. The communication sound processing device 210 includes a microphone unit 1, a band dividing unit 2, a feature amount calculating unit 3, a filter unit 4, a clarity setting unit 5, and an output unit 6. The communication control means that enables the constant connection and the means for inputting and reproducing the voice from the listener side (the other communication voice processing device) transmitted via a network (not shown) are not the main parts of the present invention. It is omitted. The microphone unit 1 collects ambient sounds and outputs the input audio signal x (n) to the band dividing unit 2 and the output unit 6 (step S1 in FIG. 2). Here, (n) indicates a discrete time discretized at a predetermined sampling interval. In FIG. 1, the A / D converter that converts the audio signal into discrete values is omitted. The band dividing unit 2 receives the input audio signal x (n) as input, performs frequency analysis by, for example, filter bank analysis, and generates a band divided signal x _m (n) divided into a predetermined frequency band (step S2). m represents the number of the divided band. For example, m = 16, and the band division signal x _m (n) divided into 16 is generated. Even if the division is not equal, division on a logarithmic scale or division according to auditory characteristics may be performed.
The greater the value of m, the closer the sound characteristics of the output signal x _c (n) output from the filter unit 4 described later will approach the input sound signal x (n). The band division unit 2 inputs the band division signal x _m (n) to the feature amount calculation unit 3.

The feature amount calculation unit 3 calculates, for example, average power for each band division signal x _m (n), and outputs a feature amount P _m of each band (step S3). The average power may be calculated by performing a smoothing process in the time direction. By smoothing, it is possible to prevent an abrupt change in the average power on the time axis. For example, assume that the current processing frame is k, the band average power in the current frame is P _tmp (k, m), and the band average power one frame before is P (k−1, m). The band average power P (k, m) may be obtained by Expression (1).

P (k, m) = (1−α) · P _tmp (k, m) + α · P (k−1, m) (1)
Here, α is a time constant for smoothing. The feature amount P _{m for} each band is input as one input signal of the filter unit 4. As the other input of the filter unit 4, an acoustic signal z (n) such as noise, environmental sound, and musical sound is input from the outside. Filter coefficient generating portion 4a of the filter 4 generates the filter coefficients D _m corresponding to the feature quantity P _m. The filter unit 4 calculates D _m * z _m (n) for each filter coefficient D _m to obtain x _c (n) (step S4). * Means a convolution operation. Here, if the acoustic signal z (n) is, for example, a bubble sound that does not feel uncomfortable even when heard for a long time, x _c (n) is an output having an average power fluctuation of the band division signal x _m (n). Become a signal. The output unit 6 receives the input audio signal x (n), the output signal x _c (n) of the filter unit 4, and the control signal c (n) and outputs an output signal y (n) (step S6). When the control signal c (n) is a value from 0 to 1, the output signal y (n) is calculated by, for example, the equation (2).

y (n) = (1-c (n)). x (n) + c (n) _.xc (n) (2)
The control signal c (n) may be a value set stepwise or a fixed value. The output signal y (n) is transmitted to a communication audio processing device on the listener side connected to the end of the network via a communication control means (not shown).

Input voice signal x (n) processed by the communication voice processing apparatus 210 of the first embodiment.
3A and 3B show signal waveforms of the output signal y (n). 3A and 3B, the horizontal axis represents time (seconds), and the vertical axis represents amplitude. FIG. 3A shows a voice waveform of a female voice in a voice database sampled at 16 kHz. Moreover, bubble sound was used for the acoustic signal z (n) not shown. FIG. 3B is a signal waveform of the output signal y (n) when the control signal c (n) = 1. The value of m is 16. In the silent section around 4, 8, 12 seconds of the input audio signal x (n), the output signal y (n) is also a silent section. When attention is paid to the shape of the time decay curve of each speech waveform, it can be seen that the waveforms of the output signal y (n) and the input speech signal x (n) are similar. In this example, since the control signal c (n) = 1, the output signal y (n) becomes the output signal x _c (n) of the filter unit 4 (formula (2)). Since the output signal is an audio signal having an average power fluctuation for each frequency band of the input audio signal x (n), the meaning of the word is unknown, but the atmosphere of the word can be transmitted. Therefore, it is possible to convey the state of talking while hiding the content of the conversation. Furthermore, since the output signal y (n) is obtained by filtering the bubble sound with the feature amount of the input sound signal, it does not feel uncomfortable even if it is heard for a long time.

  In addition, although the example which used the filter bank analysis was demonstrated as a frequency analysis means of the band division part 2, you may use short-time Fourier transform and wavelet transform. The acoustic signal z (n) has been described with an example of being supplied from the outside. However, as shown by a broken line in FIG. 1, the signal storage unit 7 is provided in the communication voice processing device 210, and the acoustic signal z (N) may be stored.

  In the first embodiment, the output signal y (n) is generated by changing the distribution of the output signal of the filter unit and the input audio signal according to the clarity set in the clarity setting unit 5. Hereinafter, this function is referred to as a privacy function for convenience of explanation. In the first embodiment, the privacy function is always operated and there is no room for selection. For example, when the communication speech processing apparatus of the present invention is used for watching an infant who does not need privacy protection, the configuration of the first embodiment may be difficult to use. A second embodiment in which the privacy function can be selectively turned on / off will be described below.

  FIG. 4 shows a functional configuration example of the communication voice processing apparatus 220 according to the second embodiment of the present invention. The operation flow is shown in FIG. The communication voice processing device 220 processes the privacy function with a frequency domain signal. In contrast to the first embodiment, in the second embodiment, a switch unit 41, a processing control unit 42, a frequency converting unit 43, and a frequency inverse converting unit 47 are added, and the operations of the band dividing unit 44 and the output unit 45 are performed. Change. Therefore, here, a newly added configuration and a portion different from the operation in the first embodiment will be described.

  The switch unit 41 is an input device having a general user interface such as various buttons, touch sensors, and dials, and outputs a switch signal p (n) to the processing control unit 42 and the output unit 45. The switch signal p (n) turns on the privacy function when p (n) = “1” (“1,0” means a logic level), and privacy function when p (n) = “0”. Is a signal to turn OFF the.

The processing control unit 42 determines the output by using the discretized input audio signal x (n) output from the microphone unit 1 and the switch signal p (n) as inputs. The processing control unit 42 inputs the input audio signal x (n) to the frequency conversion unit 43 when the switch signal p (n) = “1” (Yes in step S42 in FIG. 5). The frequency conversion unit 43 converts the input audio signal x (n) into a frequency domain signal X (ω) using, for example, a short-time Fourier transform. In the short-time Fourier transform, an input speech signal x (n) is cut out using a window function having a constant size, and the frequency domain signal X (ω) is calculated by performing fast Fourier transform (FFT) on the signal. The FFT size is a value such as 16 kHz sampling, the number of samples 256, and a shift length 1/2. The band dividing unit 44 divides the frequency domain signal X (ω) into 16 equal parts, for example, to generate a band divided signal X _m (ω) (step S44). The feature amount calculation unit 3 is the same as that in the first embodiment, and outputs the feature amount P _m of each band as one input signal of the filter unit 46. The filter unit 46 also receives externally input an acoustic frequency signal Z (ω) such as noise, environmental sound, and musical sound in the same frequency band as the frequency signal before the division of the band division signal X _m (ω). The The filter unit 46 generates Z _m (ω) obtained by dividing the acoustic frequency signal Z (ω) in accordance with the number of divisions of the feature amount P _m . Then, Z _m (ω) is filtered by the feature amount P _m (step S46). In other words, to obtain a _{X cm} (omega) to calculate the _{P m} · _Z m (ω) for each band, outputs _{X c} of the sum of the total bandwidth of _{X cm (ω) (ω)} in the frequency inverse conversion unit 47 To do. Note that the frequency band of the acoustic frequency signal Z (ω) does not necessarily match the frequency band of the input audio signal. What is necessary is just to adjust the division | segmentation number according to a desired zone | band.

The frequency inverse transform unit 47 transforms X _c (ω) output from the filter unit 46 into a time domain signal x _c (n) using, for example, short-time Fourier inverse transform (step S47).

The output unit 45 receives the control signal c (n), the switch signal p (n), the input audio signal x (n), and the output signal x _c (n) of the frequency inverse conversion unit 47 as an output signal y ( n) is synthesized and output (step S45). When the switch signal p (n) = “1”, the output unit 45 outputs the input audio signal x (n) according to the value of the control signal c (n) and the output signal x _c (n) of the frequency inverse conversion unit 47. The output signal y (n) is synthesized by changing and adding the distributions (step S451). When the switch signal p (n) = “0”, the input audio signal x (n) is output as it is as the output signal y (n) (y (n) = x (n)) (step S452, sound collection signal) Output processing). That is, the privacy function is turned off.

Thus, the privacy function can be turned ON / OFF by the switch signal p (n) output from the switch unit 41, and the usability of the communication voice processing apparatus of the present invention can be improved. The acoustic frequency signal Z (ω) is not stored as it is, but the time-domain audio data is recorded, and each time the filter unit 46 performs filtering, the audio data is subjected to frequency analysis and the acoustic frequency signal Z (ω). May be generated. Also it has been described in example of calculating the divided audio frequency signal Z (omega) in correspondence with the number of divisions of the feature quantity P _m filter unit 46 Z _{m (ω),} previously Z _m the _(omega) It may be recorded. Further, the filter operation of the filter unit 46 has been described with an example in which P _m · Z _m (ω) is calculated for each band. However, when the acoustic frequency signal Z _m (ω) is simply multiplied by the feature amount P _m. , X _c (ω) may be distorted. In that case, a coefficient W for normalization is introduced so that the average power of the entire band is equal to the average power of X _c (ω), and W · P _m · Z _m (ω) is calculated for each band. You may do it.

  The purpose of turning on / off the privacy function can also be realized by providing the switch unit 41 in the configuration of the first embodiment and changing the output unit 6 to the output unit 45.

  FIG. 6 shows a functional configuration example of the communication speech processing apparatus 230 according to the third embodiment of the present invention. The operation flow is shown in FIG. In the third embodiment, a sensor unit 61 is provided instead of the clarity setting unit 5 of the second embodiment. Other configurations are the same as those of the second embodiment. Here, operations of the sensor unit 61 and the output unit 62 will be described.

  The sensor unit 61 detects ambient environment information and outputs sensor data s (n) representing the distance between the communication voice processing device 230 and the speaker to the output unit 62 (step S61). As a detection device constituting the sensor unit 61, a microphone, a CCD image sensor, a temperature sensor, an ultrasonic sensor, an infrared sensor, or the like can be used. Any of these detection devices can be used to detect the distance between the communication speech processing apparatus 230 and the speaker with a general circuit configuration. Since the method for detecting the distance can be easily realized by the prior art, the description showing the specific configuration is omitted. The detected sensor data s (n) is converted into a value from 0 to 1 by a β (n) conversion unit 61 a provided in the sensor unit 61, for example, and input to the output unit 62.

The output unit 62 receives β (n), the input voice signal x (n), the switch signal p (n), and the output signal x _c (n) of the frequency inverse transform unit 47 as an output signal y (n). Are synthesized and output (step S62). The output unit 62 combines the output signal y (n) with the value calculated by Expression (3) (step S621).

y (n) = (1−β (n)) · x (n) + β (n) · x _c (n) (3)
The operation of the output unit 62 for the switch signal p (n) is the same as that in the second embodiment. By providing the sensor unit 61 instead of the intelligibility setting unit 5, the intelligibility of the output signal y (n) can be automatically changed according to the distance between the communication voice processing device 230 and the speaker. When the position of the speaker is close to the communication speech processing device 230, the value of β (n) is small, so that the ratio of the input speech signal x (n) to the output signal y (n) is large. Therefore, the output signal y (n) becomes clear. On the contrary, when the position of the speaker is far from the communication speech processing device 230, the ratio of the output signal x _c (n) of the filtered frequency inverse transform unit 47 becomes large, so the output signal y (n) is unclear. become. When the switch signal p (n) = “0”, the input audio signal x (n) is output as it is as the output signal y (n) (step S622).

  Thus, by providing the sensor unit 61, it becomes possible to automatically change the clarity of the output signal y (n). Note that the β (n) conversion unit 61 a may be provided in the output unit 62.

  A communication voice processing apparatus 240 according to a modification of the second embodiment is shown in FIG. The operation flow is shown in FIG. The fourth embodiment is different from the second embodiment (FIG. 4) in that the control signal c (n) set in the articulation setting unit 5 is input to the band dividing unit 81 and the operation of the output unit 82 is different. Yes. Other parts are the same as those in the second embodiment.

The band dividing unit 81 receives the frequency domain signal X (ω) from the frequency converting unit 43 and the control signal c (n) from the intelligibility setting unit 5 as input and calculates the feature amount of the band divided signal X _m (ω). It outputs to the part 3 (step S61 of FIG. 9). In the second embodiment, for example, the value of m described as m = 16 is determined by the control signal c (n). For example, when the control signal c (n) takes a value from 0 to 1, the smaller the c (n), the closer the value of m is to 32, and the larger the c (n) is, the closer m is to 16. That is, when the value of m is large and the number of band divisions is large, the bubble sound of the output signal x _c (n) of the frequency inverse transform unit 47 becomes a sound signal having characteristics closer to the input sound signal x (n). Conversely, when the value of m is small, the output signal x _c (n) of the frequency inverse transform unit 47 becomes closer to the bubble sound. The output unit 82 outputs the output signal x _c (n) of the frequency inverse transform unit 47 when the switch signal p (n) = “1”. When the switch signal p (n) = “0”, the input audio signal x (n) is output.

Therefore, when the switch signal p (n) = “1” and the value of the control signal c (n) is small, the output signal x _c (n) of the frequency inverse transform unit 47 superimposed on the output signal y (n). ) Is more like a bubble sound, the clarity of the output signal y (n) is reduced. When the value of c (n) is large, the output signal x _c (n) of the frequency inverse transform unit 47 has characteristics close to the input audio signal x (n), so the clarity of the output signal y (n) is improved. To do. As described above, even in the configuration of the fourth embodiment, the clarity of the output signal y (n) can be controlled by the value of the control signal c (n) set in the clarity setting unit 5.

  A communication voice processing apparatus 250 according to a modification of the third embodiment is shown in FIG. The operation flow is shown in FIG. The fifth embodiment is different from the third embodiment (FIG. 6) in that the sensor data s (n) output from the sensor unit 61 is input to the band dividing unit 101 and the output unit is the output unit 82. ing. The other parts are the same as in the third embodiment.

The band division unit 101 receives the frequency domain signal X (ω) from the frequency conversion unit 43 and the sensor data s (n) from the sensor unit 61 as input, and uses the band division signal X _m (ω) as the feature amount calculation unit 3. (Step S101 in FIG. 11). In this embodiment, the value of m of the band division signal X _m (ω) is determined by the sensor data s (n). The band dividing unit 101 brings the value of m closer to 32 when the value of s (n) decreases, for example, as the distance between the communication speech processing apparatus 250 and the speaker decreases, and m increases when the value of s (n) increases. To a value close to 16. The output unit 82 is the output unit 82 described in the fourth embodiment (FIG. 8). The output unit 82 synthesizes the input audio signal x (n) and the output signal x _c (n) of the frequency inverse transform unit 47 to generate an output signal. y (n) is output. If the value of m is large, the sound of the output signal x _c (n) of the frequency inverse transform unit 47 can be brought close to the input audio signal x (n). When the value of m is small, the output signal x _c (n) of the frequency inverse transform unit 47 is close to a bubble sound. Therefore, the clarity of the output signal y (n) can be controlled thereafter by the same operation as in the fourth embodiment.

  A communication voice processing apparatus 260 according to a modification of the fourth embodiment is shown in FIG. The operation flow is shown in FIG. In the sixth embodiment, the control signal c (n) output from the intelligibility setting unit 5 with respect to the fourth embodiment (FIG. 8) is input to the frequency inverse conversion unit 121 and output from the frequency conversion unit 43. The difference is that the frequency domain signal X (ω) and the acoustic frequency signal Z (ω) are input to the frequency inverse transform unit 121. Other configurations are the same as those of the fourth embodiment.

Usually, when a frequency domain signal is converted into a time domain signal by short-time inverse Fourier transform, the amplitude characteristics and phase characteristics of the frequency domain signal are used. However, the frequency inverse transform unit 121 of this embodiment varies the phase characteristic S _c (ω) of the output signal X _c (ω) of the filter unit 46 according to the control signal c (n). The frequency inverse transform unit 121 performs a short-time Fourier inverse transform on the basis of the variable phase characteristic S _c (ω) and converts the output signal X _c (ω) of the filter unit 46 into a time domain signal x _c (ω). (Step S121 in FIG. 13). The variation of the phase characteristic is an expression of the distribution of the phase characteristic S _X (ω) of the frequency domain signal X (ω) output from the frequency converter 43 and the phase characteristic S _Z (ω) of the acoustic frequency signal Z (ω). Change as shown in (4).

S _c (ω) = (1−c (n)) · S _X (ω) + c (n) · S _Z (ω) (4)
Phase characteristic S _c of the signal x _c in the time domain _{(ω) (ω),} the smaller the value of the control signal c (n), the phase characteristic S _X of the input audio signal in the frequency domain signals X _(ω) (ω) Get closer to. Conversely, if the value of c (n) is large, x _c (ω) approaches the phase characteristic S _Z (ω) of, for example, bubble sound of the acoustic frequency signal Z _m (ω). As described above, the frequency inverse transform unit 121 can also control the intelligibility of the output signal by changing the phase characteristic when the frequency domain signal is inversely transformed into the time domain signal by the control signal c (n). .

  A communication voice processing apparatus 270 according to a modification of the fourth embodiment is shown in FIG. The operation flow is shown in FIG. In the seventh embodiment, the acoustic frequency signal Z (ω) input to the filter unit 46 in the fourth embodiment (FIG. 8) is the acoustic frequency signal Za (ω) generated by the weighted addition unit 131. Is different.

  The weighted addition unit 131 receives the frequency domain signal X (ω) output from the frequency conversion unit 43, the control signal c (n) set in the articulation setting unit 5, and the acoustic frequency signal Z (ω) as inputs. An acoustic frequency signal Za (ω) is generated (step S131 in FIG. 15). For example, when the control signal c (n) takes a value from 0 to 1, the acoustic frequency signal Za (ω) is calculated by Expression (5).

Za (ω) = (1−c (n)) · X (ω) + c (n) · Z (ω) (5)
As described above, the weighted addition unit 131 weights and adds the frequency domain signal X (ω) and the acoustic frequency signal Z (ω) to generate the acoustic frequency signal Za (ω). Therefore, the acoustic frequency signal Z (ω) when the value of the control signal c (n) is small is a signal close to the input audio signal x (n). Conversely, when the value of c (n) is large, if the acoustic frequency signal Z (ω) is, for example, a bubble sound, the signal is close to a bubble sound. In this way, even with the configuration in which the weighted addition unit 131 is provided, the clarity of the output signal y (n) can be controlled.

  A communication voice processing apparatus 280 according to a modification of the third embodiment is shown in FIG. The operation flow is shown in FIG. The eighth embodiment is different from the third embodiment (FIG. 6) in that the sensor unit 61 is replaced with a power calculation unit 151 and the output unit 152.

  The power calculator 151 receives the frequency domain signal X (ω) output from the frequency converter 43 and calculates the power data Ps (n) shown in Expression (6) (step S151 in FIG. 17).

Ps (n) = Σ (X (ω) ² ) (6)
Ps (n) is the total power in one frame for which the frequency conversion unit 43 performs, for example, short-time Fourier transform. Therefore, Ps (n) is a value that increases as the amplitude of the input audio signal x (n) collected by the microphone unit 1 increases. If the value of Ps (n) is large, the distance between the communication speech processing apparatus 280 and the speaker is generally close, and conversely if the value of Ps (n) is small, the distance is far. This Ps (n) can be handled qualitatively in the same way as the sensor data of the third embodiment.

  The value of Ps (n) is input to the output unit 152 and converted to a value of 0 to 1 by the normalization unit 152a in the output unit 152. For example, the normalization unit 152a normalizes the width between the minimum value and the maximum value of Ps (n) as 1. The operation after normalizing Ps (n) is the same as that of the output unit 45 of the second embodiment. The operation of the output unit 62 for the switch signal p (n) is the same as that in the third embodiment.

  Therefore, even in the configuration in which the power calculation unit 151 is provided instead of the sensor unit 61, the intelligibility of the output signal can be automatically changed according to the distance between the communication voice processing device 280 and the speaker.

  FIG. 18 shows a functional configuration example of the communication voice processing apparatus 290 according to the ninth embodiment of the present invention. The operation flow is shown in FIG. In the ninth embodiment, the configuration of the voice section detection unit 171 and the processing control unit 42 is added to the first embodiment (FIG. 1), and the output unit is the output unit 45.

  The voice section detection unit 171 receives the input voice signal x (n) picked up by the microphone unit, determines whether or not it is a voice section, and uses the determination result d (n) as the processing control unit 42 and the output unit 45. (Step S171 in FIG. 19). The determination result d (n) is, for example, a signal in which d (n) = “1” is determined as a voice section when the short-time power of the input voice signal x (n) exceeds a threshold value for a certain time or more. . In the non-voice section, d (n) = “0”.

  The processing control unit 42 and the output unit 45 operate in the same manner as in the second embodiment by using the output signal d (n) of the voice section detection unit 171 instead of the switch signal p (n). Thus, by providing the voice section detection unit 171 instead of the switch unit 41, the privacy function can be automatically operated. In addition, Example 9 was demonstrated by the example which provides the audio | voice area detection part 171 in Example 1 (FIG. 1). The switch unit 41 of another embodiment can be replaced with a voice section detection unit 171. In that case, it becomes possible to automatically operate the privacy function only in the voice section.

  The embodiment described above is applied to communication voice processing apparatuses on the sender side and the receiver side that perform communication. Moreover, the apparatus and method which are this invention are not limited to the above-mentioned embodiment, It can change suitably in the range which does not deviate from the meaning of this invention. For example, the idea of the present invention may be applied to a network server arranged in the middle of a line. Alternatively, it may be performed only on the receiver side. In addition to the configuration of the embodiment described above, channel vocoder, formant vocoder, pattern matching vocoder, correlation vocoder, phase vocoder, maximum likelihood vocoder, homomorphic vocoder, ACOR vocoder, linear prediction using filter banks A configuration using a vocoder or LSP vocoder is also conceivable. For example, in a channel vocoder using a filter bank, the clarity is controlled by controlling the number of channels. Also, the linear prediction vocoder controls the clarity by controlling the linear prediction order. Further, the processes described in the above apparatus and method are not only executed in time series according to the order of description, but also may be executed in parallel or individually as required by the processing capability of the apparatus that executes the process. Good.

  Further, when the processing means in the above apparatus is realized by a computer, the processing contents of functions that each apparatus should have are described by a program. Then, by executing this program on the computer, the processing means in each apparatus is realized on the computer.

  The program describing the processing contents can be stored in a computer-readable storage medium. The computer-readable storage medium may be any medium such as a magnetic storage device, an optical disk, a magneto-optical storage medium, and a semiconductor memory. Specifically, for example, as a magnetic storage device, a hard disk device, a flexible disk, a magnetic tape, etc., and as an optical disk, a DVD (Digital Versatile Disc), a DVD-RAM (Random Access Memory), a CD-ROM (Compact Disc Read Only) Memory), CD-R (Recordable) / RW (ReWritable), etc., magneto-optical storage media, MO (Magneto Optical disc), etc., semiconductor memory, EEP-ROM (Electronically Erasable and Programmable-Read Only Memory), etc. Can be used.

  The program is distributed by selling, transferring, or lending a portable storage medium such as a DVD or CD-ROM storing the program, for example. Furthermore, the program may be distributed by storing the program in a storage device of the server computer and transferring the program from the server computer to another computer via a network.

  Each means may be configured by executing a predetermined program on a computer, or at least a part of these processing contents may be realized by hardware.

The figure which shows the function structural example of the communication audio | voice processing apparatus 210 of Example 1 of this invention. The figure which shows the operation | movement flow of the communication audio processing apparatus 210. The figure which shows the signal waveform of the input audio | voice signal x (n) processed by the communication audio | voice processing apparatus 210, and the output signal y (n). The figure which shows the function structural example of the communication audio processing apparatus 220 of Example 2 of this invention. The figure which shows the operation | movement flow of the communication audio processing apparatus 220. The figure which shows the function structural example of the communication audio processing apparatus 230 of Example 3 of this invention. The figure which shows the operation | movement flow of the communication audio processing apparatus 230. The figure which shows the communication audio processing apparatus 240 of Example 4 of this invention. The figure which shows the operation | movement flow of the communication audio processing apparatus 240. The figure which shows the communication audio processing apparatus 250 of Example 5 of this invention. The figure which shows the operation | movement flow of the communication audio processing apparatus 250. The figure which shows the communication audio processing apparatus 260 of Example 6 of this invention. The figure which shows the operation | movement flow of the communication audio processing apparatus 260. The figure which shows the communication audio processing apparatus 270 of Example 7 of this invention. The figure which shows the operation | movement flow of the communication audio processing apparatus 270. The figure which shows the communication audio processing apparatus 280 of Example 8 of this invention. The figure which shows the operation | movement flow of the communication audio processing apparatus 280. The figure which shows the communication audio | voice processing apparatus 290 of Example 9 of this invention. The figure which shows the operation | movement flow of the communication audio processing apparatus 290. The figure which shows the function structure of the conventional video communication apparatus 100. FIG.

Claims (7)

A frequency conversion unit for converting the input audio signal into a frequency domain signal,
And clarity setting unit for setting a bright Akirado,
The signal of the frequency domain, the intelligibility as the division number becomes large small band dividing unit for outputting a band division signals divided into a plurality of frequency bands as the number of divisions as the clarity is greater decreases When,
A feature amount calculation unit for calculating a feature amount for each of the band division signals;
For each of the plurality of frequency bands, a filter unit that multiplies a predetermined acoustic frequency signal other than the feature amount of the band- divided signal and the input audio signal ;
And frequency inverse converter for converting No. LSE out of the filter unit into a time domain signal,
Communications voice processing apparatus characterized by comprising a. A frequency conversion unit for converting the input audio signal into a frequency domain signal,
A band dividing section for outputting a band division signal by dividing the signal of the frequency domain in a predetermined frequency band,
A feature amount calculation unit for calculating a feature amount for each of the band division signals;
For each of the plurality of frequency bands, a filter unit that multiplies a predetermined acoustic frequency signal other than the feature amount of the band- divided signal and the input audio signal ;
And clarity setting unit for setting a bright Akirado,
The phase characteristic of the output signal of the filter unit is closer to the phase characteristic of the input audio signal as the clarity is higher, and closer to the phase characteristic of a predetermined acoustic frequency signal other than the input voice signal as the clarity is lower. varied as a frequency inverse conversion unit for converting the signal in the time domain,
Communications voice processing apparatus characterized by comprising a. A frequency conversion unit for converting the input audio signal into a frequency domain signal,
A band dividing section for outputting a band division signal by dividing the signal of the frequency domain in a predetermined frequency band,
A feature amount calculation unit for calculating a feature amount for each of the band division signals;
And clarity setting unit for setting a bright Akirado,
And No. LSE out of the frequency converter, and a predetermined acoustic frequency signals other than the input speech signal, the weights of the upper KiAkira Akirado becomes larger as the output signal of the frequency converter increases, the clarity is less A weighted adder that weights and adds so that the weight of the output signal of the frequency converter becomes smaller ,
For each of the plurality of frequency bands, a filter section for multiplying the output signal of the features and the weighted addition of the band division signals,
And frequency inverse converter for converting No. LSE out of the filter unit into a time domain signal,
Communications voice processing apparatus characterized by comprising a. Frequency conversion unit, a frequency conversion process of converting the input audio signal into a frequency domain signal,
Clarity setting process in which the clarity setting unit sets the clarity,
Band dividing unit, the signal of the upper Symbol frequency domain becomes large number of divided as the clarity is low, the band division signals divided into a plurality of frequency bands as the number of divisions as the clarity is greater decreases Output band splitting process,
A feature amount calculation unit calculates a feature amount for each of the band division signals, and a feature amount calculation process,
A filter process in which the filter unit multiplies the characteristic amount of the band- divided signal by a predetermined acoustic frequency signal other than the input audio signal for each of the plurality of frequency bands ;
Frequency inverse conversion unit, and the frequency inverse conversion process of converting the time domain signal No. LSE out of the filter unit,
Communications voice processing method, which comprises a. Frequency conversion unit, a frequency conversion process of converting the input audio signal into a frequency domain signal,
Band division section includes a band division step of outputting a band division signal by dividing the signal of the frequency domain in a predetermined frequency band,
A feature amount calculation unit calculates a feature amount for each of the band division signals, and a feature amount calculation process,
A filter process in which the filter unit multiplies the characteristic amount of the band- divided signal by a predetermined acoustic frequency signal other than the input audio signal for each of the plurality of frequency bands ;
Clarity setting process in which the clarity setting unit sets the clarity,
Frequency inverse conversion portion, the phase characteristic of the output signal of the filter unit, the larger the upper Symbol intelligibility close to the phase characteristic of the input speech signal, predetermined acoustic frequencies other than the input speech signal as the intelligibility is small Inverse frequency transformation process that changes to be close to the phase characteristics of the signal and converts it to a signal in the time domain,
Communications voice processing method, which comprises a. Frequency conversion unit, a frequency conversion process of converting the input audio signal into a frequency domain signal,
Band division section includes a band division step of outputting a band division signal by dividing the signal of the frequency domain in a predetermined frequency band,
A feature amount calculation unit calculates a feature amount for each of the band division signals, and a feature amount calculation process,
Clarity setting process in which the clarity setting unit sets the clarity,
Weighted addition unit, and the No. LSE out of the frequency converter, and a predetermined acoustic frequency signals other than the input speech signal, the weights of the output signals above SL clarity the larger the frequency conversion process is increased, A weighted addition process in which weighting is performed so that the weight of the output signal of the frequency converter becomes smaller as the intelligibility is smaller , and
A filtering process in which the filter unit filters the output signal of the weighted addition unit according to the feature amount for each band;
Frequency inverse conversion unit, and the frequency inverse conversion process of converting the time domain signal No. LSE out of the filter unit,
Communications voice processing method, which comprises a. A program for causing a computer to function as the communication voice processing apparatus according to any one of claims 1 to 3 .

Images