JP5216809B2 - Call quality estimation method, call quality estimation apparatus, and call quality estimation system - Google Patents

Description

  The present invention relates to a call quality estimation method, a call quality estimation apparatus, and a call quality estimation system. In particular, the present invention relates to a method for estimating the conversation quality of a service that carries out a voice call, and more particularly to a method, apparatus, and system for estimating the quality felt by a service user during a conversation from the measurement results of physical features of an evaluation target system. .

  As an existing general call quality estimation technique, there is an E-model defined in ITU-T recommendation G.107 (Non-patent Document 1). This is a technique for estimating the quality of conversation performed through an evaluation target system in consideration of the combined effect of quality parameters of the evaluation target system such as a telephone. In E-model, the parameters of the evaluation target system are digitized and the quality is estimated using an algorithm. The E-model can be used to calculate an R value that is a standard for assigning numbers to IP (Internet Protocol) telephones.

ITU-T Recommendation G.107 "The E-model, a computational model for use in transmission planning"

  The conventional comprehensive call quality estimation method including E-model assumes that both conversations experience the same conversation quality even when the conversation is biased toward one speaker. For example, it is considered that a speaker with less utterance does not affect the subjective quality as the other party has an echo amount, and a speaker with more utterance does not affect the subjective quality as the other party has voice distortion. In the conventional general call quality estimation method that assumes that both have the same conversation quality, there is a problem in that this point cannot be taken into account and the error between the quality estimate and the quality experienced by the speaker increases.

  An object of the present invention is to perform highly accurate call quality estimation reflecting a conversation form.

In order to solve the above problems, the call quality estimation method of the present invention is:
A call quality estimation method in a call quality estimation apparatus for estimating quality in an evaluation target system of a voice call service,
A transmission sound rate measurement step of measuring one transmission sound rate of the evaluation target system;
A listening sound rate measuring step for measuring one listening sound rate of the evaluation target system;
An utterance switching frequency measurement step for measuring the utterance switching frequency per unit time between transmission and listening in the evaluation target system;
A transmission psychological deterioration amount calculating step for calculating a transmission psychological deterioration amount from a quality deterioration index based on the transmission sound rate;
A listening psychological deterioration amount calculating step for calculating a listening psychological deterioration amount from a quality deterioration index based on the listening sound rate;
A delayed psychological deterioration amount calculating step for calculating a delayed psychological deterioration amount from a quality deterioration index based on the number of utterance switching times;
An integrated psychological deterioration amount calculating step for calculating an integrated psychological deterioration amount from the transmitted psychological deterioration amount, the listening psychological deterioration amount, and the delayed psychological deterioration amount;
A subjective quality evaluation value calculating step for converting the integrated psychological deterioration amount into a subjective quality evaluation value;
It is characterized by having.

In addition, the call quality estimation device of the present invention,
A call quality estimation device for estimating quality in an evaluation target system of a voice call service,
A transmission sound rate measurement unit for measuring one transmission sound rate of the evaluation target system;
A listening sound rate measuring unit for measuring one listening sound rate of the evaluation target system;
An utterance switching number measuring unit for measuring the number of utterance switching per unit time between transmission and listening in the evaluation target system;
A transmission psychological deterioration amount calculation unit that calculates a transmission psychological deterioration amount from a quality deterioration index based on the transmission sound rate;
A listening psychological deterioration amount calculating unit for calculating a listening psychological deterioration amount from a quality deterioration index based on the listening sound rate;
A delayed psychological deterioration amount calculating unit for calculating a delayed psychological deterioration amount from a quality deterioration index based on the number of utterance switching times;
An integrated psychological deterioration amount calculating unit that calculates an integrated psychological deterioration amount from the transmitted psychological deterioration amount, the listening psychological deterioration amount, and the delayed psychological deterioration amount;
A subjective quality evaluation value calculation unit for converting the integrated psychological deterioration amount into a subjective quality evaluation value;
It is characterized by having.

The call quality estimation system of the present invention is
A call quality estimation system for estimating the quality of a voice call service evaluation target system,
A first quality deterioration index calculating device connected to one of the evaluation target systems and measuring one quality deterioration index of the evaluation target system;
A second quality deterioration index calculating device connected to the other of the evaluation target system and measuring the other quality deterioration index of the evaluation target system;
A subjective quality evaluation value calculating device connected to the first quality deterioration index calculating device and the second quality deterioration index calculating device;
Have
Subjective quality evaluation value calculation device
A transmission sound rate measurement unit for measuring one transmission sound rate of the evaluation target system;
A listening sound rate measuring unit for measuring one listening sound rate of the evaluation target system;
An utterance switching number measuring unit for measuring the number of utterance switching per unit time between transmission and listening in the evaluation target system;
Based on the transmission sound rate and the listening sound rate, a transmission psychological deterioration amount calculation unit that calculates a transmission psychological deterioration amount from a measured quality deterioration index;
Based on the transmission sound rate and the listening sound rate, a listening psychological deterioration amount calculating unit that calculates a listening psychological deterioration amount from a measured quality deterioration index;
A delayed psychological deterioration amount calculation unit that calculates a delayed psychological deterioration amount from the measured quality deterioration index based on the number of utterance switching times;
An integrated psychological deterioration amount calculating unit that calculates an integrated psychological deterioration amount from the transmitted psychological deterioration amount, the listening psychological deterioration amount, and the delayed psychological deterioration amount;
A subjective quality evaluation value calculation unit for converting the integrated psychological deterioration amount into a subjective quality evaluation value;
It is characterized by having.

  According to the present invention, it is possible to perform highly accurate call quality estimation reflecting the conversation form.

The block diagram which shows the structure of the speech quality estimation apparatus which concerns on Embodiment 1 of this invention. The block diagram which shows the structure of the call quality estimation system which concerns on Embodiment 2 of this invention.

  In the present invention, in an evaluation target system such as an IP phone, the call quality is estimated based on the speech rate (transmission speech rate and listening speech rate) of conversations exchanged by a service for performing a voice call, and the number of speech switchings. . Specifically, a quality degradation index (delay time, voice distortion and echo amount) caused by the evaluation target system is measured. Next, measure the speech rate of one of the evaluation target systems, measure the listening rate of one of the evaluation target systems, and switch the utterance per unit time of transmission and reception in the evaluation target system Measure the number of times. Based on the measured speech rate, the transmission psychological degradation amount is calculated from the quality degradation index (delay time and echo amount), and listening from the quality degradation index (sound distortion) based on the measured listening rate A psychological deterioration amount is calculated, and a delayed psychological deterioration amount is calculated from a quality deterioration index (delay time) based on the measured number of utterance changes. An integrated psychological deterioration amount is calculated from the calculated transmitted psychological deterioration amount, listening psychological deterioration amount, and delayed psychological deterioration amount, and the calculated integrated psychological deterioration amount is converted into a subjective quality evaluation value.

  As described above, the speech rate and the number of utterance changes in the conversation with respect to the transmitted psychological degradation amount, the listening psychological degradation amount, and the delayed psychological degradation amount measured from the quality degradation index (value of quality degradation factor) generated during the conversation. By assigning the corresponding weights, it is possible to estimate the total user experience call quality more accurately than in the conventional method.

  Hereinafter, embodiments of the present invention will be described in detail.

(Embodiment 1)
Embodiment 1 below estimates the overall call quality on one side of an evaluation target system such as an IP phone, and the quality degradation indicators for delay time, echo amount, encoding method and packet loss rate are given in advance. Assuming that

  Embodiment 1 of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram illustrating a configuration of a call quality estimation apparatus 11 according to the first embodiment. The call quality estimation device 11 includes a transmission sound rate measurement unit 12, a listening sound rate measurement unit 13, an utterance switching frequency measurement unit 14, a transmission psychological deterioration amount calculation unit 15, a listening psychological deterioration amount calculation unit 16, a delayed psychology. A deterioration amount calculation unit 17, an integrated psychological deterioration amount calculation unit 18, and a subjective quality evaluation value calculation unit 19 are provided.

  As a quality degradation index due to the evaluation target system 100 on one side of the conversation, the delay time D [ms] is sent to the sentiment deterioration deterioration amount calculation unit 15 and the delayed mind deterioration amount calculation unit 17, and the echo amount E [dB] is sent to the speech. The psychological degradation amount calculation unit 15 is provided with the encoding method and the packet loss rate P [%] as input to the listening psychological degradation amount calculation unit 16.

  The transmission sound rate measurement unit 12 receives a transmission voice signal S of t × x [ms] via the evaluation target system 100, and the transmission voice signal sound rate (transmission sound rate) As ( Measure 0 ≦ As ≦ 1). A method for measuring the sound rate of the transmitted voice signal is shown below. The signal S is divided into x signals Si (1 ≦ i ≦ x) every t [ms], and the number y of signals whose volume exceeds the threshold Ls [dBov] among the x signals Si is calculated. Then, As = y / x is calculated and transmitted to the transmission psychological deterioration amount calculation unit 15 as the transmission sound rate As. As described above, the transmission sound rate As can be measured by analyzing the transmission signal per fixed time. Here, t = 200, x = 300, and Ls = −46, but are not limited thereto.

  The listening sound rate measuring unit 13 receives the t × x [ms] listening sound signal R via the evaluation target system 100, and the sound rate of the listening sound signal (listening sound rate) Ar (0 ≦ Ar ≦). 1) Measure. A method for measuring the sound rate of the listening audio signal is shown below. The signal R is divided into x signals Ri (1 ≦ i ≦ x) every t [ms], and the number z of signals whose volume exceeds the threshold Lr [dBov] among the x signals Ri is calculated. Then, Ar = z / x is calculated and transmitted to the listening psychological deterioration amount calculation unit 16 as a listening sound rate Ar. In this way, the listening sound rate Ar can be measured by analyzing the listening signal per certain time. Here, Lr = −46, but it is not limited to this.

  The utterance switching frequency measuring unit 14 receives the transmission voice signal S of t × x [ms] and the listening voice signal R of t × x [ms] via the evaluation target system 100, and a unit of transmission and listening. Measure the number of utterance changes B per hour. The method for measuring the number of utterance changes is shown below. The signals S and R are divided into x signals Si and Ri (1 ≦ i ≦ x) every t seconds. Then, the number Bs of sections in which M or more signals Si all exceed the threshold value Ls with respect to the signal Si, and there are N or more consecutive signal Si signals that are below the threshold value immediately before and after that, is detected. To do. Similarly, the number of sections Br in which M or more signals Ri exceed the threshold Lr, and there are N or more consecutive signals Ri below the threshold immediately before and immediately after the signal Ri, respectively. To do. In this way, the transmission utterance count Bs and the listening utterance count Br are measured. Then, B = 2 × min (Bs, Br) is calculated, and this B is transmitted to the delayed psychological deterioration amount calculation unit 17 as the number of utterance switching B. Thus, by analyzing the transmission signal and the listening signal per certain time, the number of utterance switching B can be measured. Here, although M = 10 and N = 10, it is not limited to this.

  The transmission psychological deterioration amount calculation unit 15 receives the transmission sound rate As, the delay time D, and the echo amount E as inputs. For example, based on the relationship between the echo amount and delay time and the subjective evaluation value for quality (average opinion evaluation value MOS defined in ITU-T recommendation P.800), E-T proposed by ITU-T recommendation G.107 In the model, the degradation amount Idte due to the talker echo is defined as the following equations (1) to (4).

ここで、式(1), (4)内のT, TELRはそれぞれ片道遅延時間、エコー量に対応しているため、それぞれにD, Eを入力する。さらにNo, RLRにはデフォルト値である-61.18, 2をそれぞれ式(2)に代入しIdteを算出する。この送話者エコーによる劣化量Idteは送話者のみが知覚する劣化であるため、送話信号の有音率Asが高いほど送話心理劣化量は増大することになる。従って、送話心理劣化量Gsは、品質劣化指標から送話品質劣化量への変換式をあらかじめ定め、この変換式により算出される送話品質劣化量に対して、送話有音率Asに基づいた重み付けをすることにより算出されてもよい。送話有音率Asに基づいて品質劣化指標から送話心理劣化量Gsを算出するために、Idte, As, Gsの関係をAsの増加につれGsが増加するような単調増加関数で表し、前記関数を用いてGsを算出する。ここでは単調増加関数を式(5)で表しているが、これに限定されるものではない。式(5)のa1, a2は定数である。

Here, since T and TELR in the equations (1) and (4) correspond to the one-way delay time and the echo amount, D and E are input to each. Furthermore, Idte is calculated by substituting the default values -61.18 and 2 into No. and RLR, respectively, in equation (2). Since the deterioration amount Idte due to the transmitter echo is a deterioration perceived only by the transmitter, the transmission psychological deterioration amount increases as the sound rate As of the transmission signal increases. Therefore, the transmission psychological degradation amount Gs is determined in advance by a conversion formula from the quality degradation index to the transmission quality degradation amount, and the transmission speech deterioration rate As is calculated with respect to the transmission quality degradation amount calculated by this conversion formula. It may be calculated by weighting based on it. In order to calculate the transmission psychological deterioration amount Gs from the quality deterioration index based on the transmission sound rate As, the relationship between Idte, As, Gs is expressed by a monotonically increasing function such that Gs increases as As increases, Gs is calculated using the function. Here, the monotonically increasing function is represented by Expression (5), but the present invention is not limited to this. In formula (5), a1 and a2 are constants.

このようにして算出した送話心理劣化量Gsを統合心理劣化量算出部１８に送信する。

The transmission psychological deterioration amount Gs thus calculated is transmitted to the integrated psychological deterioration amount calculation unit 18.

  The listening psychological deterioration amount calculation unit 16 receives the listening sound rate Ar, the encoding method, and the packet loss rate P as inputs. For example, from the relationship between the coding method and the packet loss rate and the subjective evaluation value of the quality, the distortion degradation amount Ie, eff due to coding distortion and packet loss in the E-model proposed by ITU-T recommendation G.107 is It is defined as the following equation (6).

ここで、式(6)内のPplはパケット損失率に対応しているため、Pを入力する。さらにIe, Bplはそれぞれ符号化による品質劣化量、符号化方式のパケット損失耐性を表しており、ITU-T勧告G.113 Appendix Iは各符号化方式のIeおよびBplの値を示している。そこで、入力された符号化方式情報に基づきIe, Bplを決定し、式(6)に代入する。BurstRにはデフォルト値である1を式(6)に代入し、Ie,effを算出する。この歪み劣化量Ie,effは受聴者のみが知覚する劣化であるため、受聴信号の有音率Arが高いほど受聴心理劣化量は増大することになる。従って、受聴心理劣化量Grは、品質劣化指標から受聴品質劣化量への変換式をあらかじめ定め、この変換式により算出される受聴品質劣化量に対して、受聴有音率Arに基づいた重み付けをすることにより算出されてもよい。受聴有音率Arに基づいて品質劣化指標から受聴心理劣化量Grを算出するために、Ie,eff, Ar, Grの関係をArの増加につれGrが増加するような単調増加関数で表し、前記関数を用いてGrを算出する。ここでは単調増加関数を式(7)で表しているが、これに限定されるものではない。式(7)のa3, a4は定数である。

Here, Ppl in Expression (6) corresponds to the packet loss rate, so P is input. Further, Ie and Bpl represent the quality degradation amount due to encoding and the packet loss tolerance of the encoding method, respectively, and ITU-T recommendation G.113 Appendix I indicates the values of Ie and Bpl of each encoding method. Therefore, Ie and Bpl are determined based on the input coding method information, and are substituted into equation (6). For BurstR, the default value of 1 is substituted into Equation (6), and Ie and eff are calculated. Since this distortion deterioration amount Ie, eff is deterioration perceived only by the listener, the listening psychological deterioration amount increases as the sound rate Ar of the listening signal increases. Therefore, the listening psychological degradation amount Gr is determined in advance by a conversion formula from the quality degradation index to the listening quality degradation amount, and the listening quality degradation amount calculated by this conversion formula is weighted based on the listening sound rate Ar. May be calculated. In order to calculate the listening psychological deterioration amount Gr from the quality deterioration index based on the listening sound rate Ar, the relationship between Ie, eff, Ar, Gr is expressed by a monotonically increasing function such that Gr increases as Ar increases, Gr is calculated using the function. Here, the monotonically increasing function is represented by Expression (7), but the present invention is not limited to this. In formula (7), a3 and a4 are constants.

このようにして算出した受聴心理劣化量Grを統合心理劣化量算出部１８に送信する。

The listening psychological deterioration amount Gr calculated in this way is transmitted to the integrated psychological deterioration amount calculation unit 18.

  In the delayed psychological deterioration amount calculation unit 17, the number of utterance switching B and the delay time D are given as inputs. For example, based on the relationship between the delay time and the subjective evaluation value of the quality, the degradation amount Idd due to delay in the E-model proposed by ITU-T recommendation G.107 is defined as in the following equations (8)-(10) Has been.

ここで、式(8)-(10)のTaは片道遅延時間に対応しているため、Dを入力しIddを算出する。この遅延劣化量Iddは発話切り替わり時に特に知覚する劣化であるため、発話切り替わり回数Bが高いほど遅延心理劣化量は増大することになる。従って、遅延心理劣化量Gdは、品質劣化指標から遅延品質劣化量への変換式をあらかじめ定め、この変換式により算出される遅延品質劣化量に対して、発話切り替わり回数Bに基づいた重み付けをすることにより算出されてもよい。発話切り替わり回数Bに基づいて品質劣化指標から遅延心理劣化量Gdを算出するために、Idd, B, Gdの関係をBの増加につれGdが増加するような単調増加関数で表し、前記関数を用いてGdを算出する。ここでは単調増加関数を式(11)で表しているが、これに限定されるものではない。式(11)のa5, a6は定数である。

Here, since Ta in equations (8)-(10) corresponds to the one-way delay time, D is input to calculate Idd. Since this delay deterioration amount Idd is a deterioration that is particularly perceived at the time of utterance switching, the delay psychological deterioration amount increases as the number of utterance switching times B increases. Accordingly, the delay psychological deterioration amount Gd is determined in advance by a conversion formula from the quality deterioration index to the delay quality deterioration amount, and the delay quality deterioration amount calculated by this conversion formula is weighted based on the number of utterance switching B. May be calculated. In order to calculate the delayed psychological deterioration amount Gd from the quality deterioration index based on the number of utterance switching B, the relationship between Idd, B, Gd is expressed by a monotonically increasing function such that Gd increases as B increases, and the above function is used. To calculate Gd. Here, the monotonically increasing function is represented by Expression (11), but the present invention is not limited to this. In formula (11), a5 and a6 are constants.

このようにして算出した遅延心理劣化量Gdを統合心理劣化量算出部１８に送信する。

The delayed psychological deterioration amount Gd calculated in this way is transmitted to the integrated psychological deterioration amount calculation unit 18.

  In the integrated psychological deterioration amount calculation unit 18, the transmission psychological deterioration amount Gs, the listening psychological deterioration amount Gr, and the delayed psychological deterioration amount Gd are given as inputs. The cumulative value Gt (= Gs + Gr + Gd) of Gs, Gr, and Gd is set as the integrated psychological deterioration amount, and the integrated psychological deterioration amount Gt is transmitted to the subjective quality evaluation value calculation unit 19.

  In the subjective quality evaluation value calculation unit 19, the integrated psychological deterioration amount Gt is given as an input. In the E-model proposed by ITU-T recommendation G.107, a value obtained by subtracting the integrated psychological deterioration amount Gt from the non-deteriorating default value 93.2 of the non-degrading quality scale value R value is used as the evaluation target system 100. The quality measure value R is converted into the subjective quality evaluation value MOS by the following equations (12) and (13) defined in ITU-T recommendation G.107 Appendix I.

このようにして得られたMOSを評価対象系１００の主観品質評価値として出力する。

The MOS obtained in this way is output as the subjective quality evaluation value of the evaluation target system 100.

  The procedure for formulating a monotonically increasing function that represents weighting based on the sound rate and the number of switching is shown below.

  It shows about weighting of listening psychological deterioration. First, a plurality of experimental conditions are set for speech distortion, and a one-minute conversation is performed. Conversation opinion evaluation experiments based on ITU-T recommendation P.800 will be conducted. Furthermore, for each experimental condition, a conversation task with a different listening sound rate Ar is given. Next, the MOS values for all experimental conditions are converted into R values by the above-described method, and the listening psychological degradation amount Gr subtracted from the reference value 93.2 is calculated. Then, the distortion deterioration amount Ie, eff in each experimental condition is measured, and the weighting Wr (= Gr / Ie, eff) in each listening sound rate Ar is calculated using the listening psychological deterioration amount Gr in that condition. Finally, a3 and a4 that minimize the mean square error of the monotonically increasing function Wr = a3Ar + a4 are calculated by the method of least squares. When a non-linear function is used for the monotonically increasing function, a3 and a4 are calculated using a non-linear least squares fitting mechanism. At this time, a3 and a4 may be divided by the same value so that Wr becomes 1 when the audible sound rate Ar is a sound rate determined as a reference (for example, 0.4).

  The weighting of the delayed psychological deterioration will be described. First, a plurality of experimental conditions are set for the delay time, and a one-minute conversation is conducted. Conversation opinion evaluation experiments based on ITU-T recommendation P.800 will be conducted. Furthermore, for each experimental condition, a conversation task with different utterance switching times B is given. Next, the MOS values for all the experimental conditions are converted to R values by the above-described method, and the delayed psychological deterioration amount Gd subtracted from the reference value 93.2 is calculated. Then, the deterioration amount Idd due to delay under each experimental condition is measured, and the weight Wd (= Gd / Idd) at the number of utterance switching B is calculated using the delayed psychological deterioration amount Gd under that condition. Finally, a5 and a6 at which the mean square error of the monotonically increasing function Wd = a5B + a6 is minimized are calculated by the least square method calculation. When a non-linear function is used for the monotonically increasing function, a5 and a6 are calculated using a non-linear least squares fitting mechanism. At this time, a5 and a6 may be divided by the same value so that Wd becomes 1 when the number of utterance switching B is a reference number (for example, 10).

  The weighting of the sentiment deterioration amount will be described. First, a plurality of experimental conditions are set for the echo amount and the delay time, and a one-minute conversation is performed. Conversation opinion evaluation experiments based on ITU-T recommendation P.800 will be conducted. Furthermore, for each experimental condition, a conversation task with a different transmission sound rate As is given. Next, the MOS values for all experimental conditions are converted to R values by the above method, and the value obtained by subtracting the calculated value of the delayed psychological deterioration amount by the method described above from the value obtained by subtracting from the reference value 93.2 Is calculated as the transmission psychological deterioration amount Gs. Then, the deterioration amount Idte due to the transmitter echo under each experimental condition is measured, and the weighting Ws (= Gs / Idte) in the transmission sound rate As is calculated using the transmission psychological deterioration amount Gs under the condition. Finally, a1 and a2 that minimize the mean square error of the monotonically increasing function Ws = a1As + a2 are calculated by the least square method calculation. When a non-linear function is used for the monotonically increasing function, a1 and a2 are calculated using a non-linear least squares fitting mechanism. At this time, a1 and a2 may be divided by the same value so that Ws becomes 1 when the transmission sound rate As is a sound rate determined as a reference (for example, 0.4).

(Embodiment 2)
In the second embodiment, one total call quality is estimated based on the total call quality on both sides of an evaluation target system such as an IP phone, and the quality deterioration index of delay time, echo amount, and voice distortion is the evaluation target system. Measure directly from In the second embodiment, a test signal is transmitted / received via the evaluation target system, and a quality deterioration index is measured.

  Embodiment 2 of the present invention will be described with reference to the drawings. FIG. 2 is a block diagram illustrating a configuration of the call quality estimation system 20 according to the second embodiment. The call quality estimation system 20 includes a subjective quality evaluation value calculation device 21, a quality deterioration index calculation device 31, and a quality deterioration index calculation device 41.

  The subjective quality evaluation value calculation device 21 includes a transmission sound rate measurement unit 22, a listening sound rate measurement unit 23, a speech switching frequency measurement unit 24, a transmission psychological deterioration amount calculation unit 25, a listening psychological deterioration amount calculation unit 26, A delayed psychological deterioration amount calculation unit 27, an integrated psychological deterioration amount calculation unit 28, and a subjective quality evaluation value calculation unit 29 are provided.

  The quality deterioration index calculation unit 31 is connected to one side of the evaluation target system, and calculates a quality deterioration index on one side of the evaluation target system. The quality degradation index calculation unit 31 includes a signal generation unit 32, a delay time measurement unit 33, an echo amount measurement unit 34, and a distortion amount measurement unit 35.

  The quality deterioration index calculation unit 41 is connected to the opposite side of the evaluation target system, and calculates a quality deterioration index on the opposite side of the evaluation target system. The quality degradation index calculation unit 41 includes a signal generation unit 42, an echo amount measurement unit 44, and a distortion amount measurement unit 45.

  The transmission voice rate measurement unit 22 receives the transmission voice signal S of t × x [ms] via the evaluation target system 101, and the voice rate of the transmission voice signal (transmission voice rate) As ( Measure 0 ≦ As ≦ 1). The sound rate As of the transmitted voice signal is transmitted to the transmitted psychological deterioration amount calculating unit 25 and the listening psychological deterioration amount calculating unit 26. The method for measuring the sound rate of the transmitted voice signal is the same as in the first embodiment. Here, t = 200 and x = 300 are set, but not limited thereto.

  The listening sound rate measuring unit 23 receives the t × x [ms] listening sound signal R via the evaluation target system 101, and the sound rate (listening sound rate) Ar (0 ≦ Ar ≦) of the listening sound signal. 1) Measure. The sound rate Ar of the listening voice signal is transmitted to the transmission psychological deterioration amount calculating unit 25 and the listening psychological deterioration amount calculating unit 26. The method for measuring the sound rate of the listening audio signal is the same as in the first embodiment.

  The utterance switching number measuring unit 24 receives the t × x [ms] transmission voice signal S and the t × x [ms] listening voice signal R via the evaluation target system 101 and measures the utterance switching number B. . This utterance switching frequency B is transmitted to the delayed psychological deterioration amount calculation unit 27. The method for measuring the number of utterance switching is the same as in the first embodiment.

  An audio signal is stored in the signal generation unit 32, and the audio signal S1 is transmitted to the evaluation target system 101 when an instruction for signal generation is received from the echo amount measurement unit 34 or the distortion amount measurement unit 35. At this time, the audio signal S1 is transmitted to the echo amount measuring unit 34 or the distortion amount measuring unit 35 that has given the instruction.

  The delay time measuring unit 33 calculates a round trip delay time D ′ [ms] between the two by transmitting a Ping (Packet InterNet Groper) from the receiving side to the transmitting side, and a one-way delay time D = D ′ / 2 is calculated. This one-way delay time D is transmitted to the transmission psychological deterioration amount calculation unit 25 and the delayed psychological deterioration amount calculation unit 27, respectively.

  In the echo amount measurement unit 34, the signal generation unit 32 is instructed to transmit the audio signal S1 to one side of the evaluation target system 101, and the audio signal S1 from the signal generation unit 32 is the same through the evaluation target system. The audio signal S2 to be heard on the side is acquired. Then, the level difference E1 [dB] between the audio signal S1 and the listening audio signal S2 is calculated, and the measured echo amount E1 [dB] is transmitted to the transmission psychological deterioration amount calculation unit 25. The echo amount E1 is an echo amount received on the side where the transmission signal S and the listening signal R are acquired.

  In the distortion amount measurement unit 35, the signal generation unit 32 is instructed to transmit the audio signal S1 to one side of the evaluation target system 101, the audio signal S1 is transmitted from the signal generation unit 32, and the evaluation target system 101 is opposite to the evaluation target system. The audio signal S3 to be heard on the side is acquired. The voice quality evaluation value PESQ is calculated from the voice signals S1 and S3 by using the PESQ of ITU-T recommendation P.862. This PESQ is converted into an R value (R) by the following equations (14) to (17) defined in ITU-T recommendation G.107 Appendix I.

同様に音声信号S1を用いて同一信号の比較による音声品質評価値PESQを計算し、式(14)-(17)によりR値(R')に変換する。２つのR値の差分Ie,eff（=R'-R）を音声信号の歪み量Ie,eff1として算出し、測定された歪み量Ie,eff1を受聴心理劣化量算出部２６に送信する。なお、この歪み量Ie,eff1は、送話信号Sおよび受聴信号Rを取得した側で受聴する歪み量である。

Similarly, a voice quality evaluation value PESQ based on comparison of the same signal is calculated using the voice signal S1, and is converted into an R value (R ′) according to equations (14)-(17). The difference Ie, eff (= R′−R) between the two R values is calculated as the distortion amount Ie, eff1 of the audio signal, and the measured distortion amount Ie, eff1 is transmitted to the listening psychological deterioration amount calculation unit 26. The distortion amounts Ie and eff1 are distortion amounts received on the side where the transmission signal S and the listening signal R are acquired.

  Similarly, an audio signal is stored in the signal generation unit 42, and when an instruction for signal generation is received from the echo amount measurement unit 44 or the distortion amount measurement unit 45, the audio signal S1 ′ is sent to the evaluation target system 101. Send. At this time, the audio signal S1 ′ is transmitted to the echo amount measuring unit 44 or the distortion amount measuring unit 45 that has given the instruction.

  Note that the delay time is assumed to obtain the same result regardless of which Ping is transmitted from, so the quality degradation index calculation device 41 may not include the delay time measurement unit.

  The echo amount measurement unit 44 instructs the signal generation unit 42 to transmit the audio signal S1 ′ to the opposite side of the evaluation target system 101 (the side opposite to the side from which the transmission signal S and the listening signal R are acquired). The sound signal S1 ′ is acquired from the generation unit 42, and the sound signal S2 ′ received from the evaluation target system 101 through the evaluation target system on the same side (the side opposite to the side from which the transmission signal S and the listening signal R are acquired) is acquired. . Similarly to the echo amount measuring unit 34, the level difference E2 [dB] between the audio signal S1 'and the listening audio signal S2' is calculated, and the measured echo amount E2 [dB] is transmitted to the transmission psychological deterioration amount calculating unit 25. To do.

  The distortion amount measurement unit 45 instructs the signal generation unit 42 to transmit the audio signal S1 ′ to the opposite side of the evaluation target system 101 (the side opposite to the side from which the transmission signal S and the listening signal R have been acquired). The sound signal S1 ′ is acquired from the generation unit 42, and the sound signal S3 ′ received from the evaluation target system 101 on the opposite side of the evaluation target system (the side from which the transmission signal S and the listening signal R are acquired) is acquired. Similar to the distortion amount measurement unit 35, the difference Ie, eff (= R′−R) between the two R values is calculated as the distortion amount Ie, eff2 of the audio signal, and the measured distortion amount Ie, eff2 is deteriorated in listening psychology. It transmits to the quantity calculation part 26.

  In the transmission psychological deterioration amount calculation unit 25, a transmission sound rate As, a listening sound rate Ar, a delay time D, and echo amounts E1 and E2 are given as inputs. The transmitted psychological deterioration amounts Gs1, Gs2 calculated from these are transmitted to the integrated psychological deterioration amount calculation unit 28. Gs1 is the transmission psychological deterioration amount experienced on the side where the transmission signal S and the listening signal R are acquired. Gs1 is Gs calculated by using E1 instead of E in the calculation method in the first embodiment. Gs2 is Gs calculated by using Ar and E2 instead of As and E, respectively, in the calculation method in the first embodiment. In other words, the transmission psychological degradation amount Gs2 on the opposite side is calculated from the listening sound rate Ar on the side that acquired the transmission signal S and the listening signal R. Since this listening sound rate Ar can be regarded as the transmission sound rate on the opposite side, instead of the listening sound rate Ar on the side that acquired the transmission signal S and the listening signal R, the transmission rate on the opposite side The sound rate may be used.

  The listening psychological deterioration amount calculation unit 26 receives the transmission sound rate As, the listening sound rate Ar, and the distortion amounts Ie, eff1, Ie, and eff2 as inputs. The listening psychological deterioration amounts Gr1, Gr2 calculated from these are transmitted to the integrated psychological deterioration amount calculation unit 28. This is the listening psychological degradation amount that Gr1 experiences on the side where the transmission signal S and the listening signal R are acquired. Gr1 is Gr calculated by using Ie, eff1 instead of Ie, eff in the calculation method in the first embodiment. Gr2 is Gr calculated by using As, Ie, and eff2 in place of Ar, Ie, and eff in the calculation method according to the first embodiment. That is, the listening psychological degradation amount Gr2 on the opposite side is calculated from the transmission sound rate As on the side from which the transmission signal S and the listening signal R are acquired. Since this transmission sound rate As can be regarded as the listening sound rate on the opposite side, instead of the transmission sound rate As on the side that acquired the transmission signal S and the listening signal R, the listening sound rate on the opposite side The sound rate may be used.

  In the delayed psychological deterioration amount calculation unit 27, the number of utterance switching B and the delay time D are given as inputs. The delayed psychological deterioration amount Gd calculated from these is transmitted to the integrated psychological deterioration amount calculation unit 28. The method for calculating the sentiment deterioration amount Gd is the same as that in the first embodiment.

  The integrated psychological deterioration amount calculation unit 28 is provided with input psychological deterioration amounts Gs1, Gs2, listening psychological deterioration amounts Gr1, Gr2, and delayed psychological deterioration amounts Gd1, Gd2. Here, the integrated psychological deterioration amount Gt is calculated by the equation (18), assuming that the worst quality of the bodily sensation quality at both ends is output. Although the worst value is taken here, it may be an average value or a best value.

この統合心理劣化量Gtを主観品質評価値算出部２９に送信する。

This integrated psychological deterioration amount Gt is transmitted to the subjective quality evaluation value calculation unit 29.

  The subjective quality evaluation value calculation unit 29 is the same as that in the first embodiment.

(Effect of embodiment)
As described above, according to the comprehensive call quality estimation method of the present invention, it is possible to estimate call quality with high accuracy reflecting the conversation form. More specifically, it is possible to estimate the call quality taking into account the conversation bias that there are few utterances.

  Thus, since the user experience call quality in consideration of the conversation form can be estimated, it is possible to estimate the total call quality that captures the call service form and conversation characteristics. In addition, by establishing a standard of the utterance ratio at both ends in the service provided in advance, by executing the call quality estimation according to the present invention at both ends, it is possible to perform an appropriate quality design considering the quality of experience of both users.

  For convenience of explanation, the device and system according to the embodiments of the present invention are described using functional block diagrams. However, the device and system of the present invention may be realized by hardware, software, or a combination thereof. Good. For example, each functional unit of the device and the system may be realized by software. In addition, the functional units may be used in combination as necessary.

  Although the embodiments of the present invention have been described above, the present invention is not limited to the above-described embodiments, and various modifications and applications are possible within the scope of the claims.

  The present invention can be applied to quality design in consideration of, for example, a conversation form assumed in an IP telephone service.

100, 101 Evaluation target system 11 Call quality estimation device 20 Call quality estimation system 21 Subjective quality evaluation calculation device 12, 22 Transmission sound rate measurement unit 13, 23 Listen sound rate measurement unit 14, 24 Speech switching frequency measurement unit 15 25 Psychological degradation amount calculation unit 16, 26 Listening psychological degradation amount calculation unit 17, 27 Delayed psychological degradation amount calculation unit 18, 28 Total psychological degradation amount calculation unit 19, 29 Subjective quality evaluation value calculation unit 31, 41 Quality degradation Index calculation device 32, 42 Signal generation unit 33 Delay time measurement unit 34, 44 Echo amount measurement unit 35, 45 Distortion amount measurement unit

Claims (10)

A call quality estimation method in a call quality estimation apparatus for estimating quality in an evaluation target system of a voice call service,
A transmission sound rate measurement step of measuring one transmission sound rate of the evaluation target system;
A listening sound rate measuring step for measuring one listening sound rate of the evaluation target system;
An utterance switching frequency measurement step for measuring the utterance switching frequency per unit time between transmission and listening in the evaluation target system;
A transmission psychological deterioration amount calculating step for calculating a transmission psychological deterioration amount from a quality deterioration index based on the transmission sound rate;
A listening psychological deterioration amount calculating step for calculating a listening psychological deterioration amount from a quality deterioration index based on the listening sound rate;
A delayed psychological deterioration amount calculating step for calculating a delayed psychological deterioration amount from a quality deterioration index based on the number of utterance switching times;
An integrated psychological deterioration amount calculating step for calculating an integrated psychological deterioration amount from the transmitted psychological deterioration amount, the listening psychological deterioration amount, and the delayed psychological deterioration amount;
A subjective quality evaluation value calculating step for converting the integrated psychological deterioration amount into a subjective quality evaluation value;
A call quality estimation method comprising: The transmission psychological deterioration amount calculating step calculates a transmission psychological deterioration amount from a delay time and an echo amount based on the transmission sound rate,
The listening psychological deterioration amount calculating step calculates a listening psychological deterioration amount from sound distortion based on the listening sound rate,
The call quality estimation method according to claim 1, wherein the delay psychological deterioration amount calculating step calculates a delay psychological deterioration amount from a delay time based on the number of utterance switching. A first quality deterioration index calculating step in which a first quality deterioration index calculating device connected to one of the evaluation target systems measures one quality deterioration index of the evaluation target system;
A second quality deterioration index calculating step in which a second quality deterioration index calculating device connected to the other evaluation target system measures the other quality deterioration index of the evaluation target system;
Further comprising
The transmission psychological deterioration amount calculating step calculates a transmission psychological deterioration amount from a measured quality deterioration index based on the transmission sound rate and the listening sound rate,
The telephone conversation according to claim 1 or 2, wherein the listening psychological deterioration amount calculating step calculates a listening psychological deterioration amount from a measured quality deterioration index based on the transmission sound rate and the listening sound rate. Quality estimation method. The first quality deterioration index calculating step includes:
A first signal generating step for generating a first audio signal;
A delay time measuring step for measuring the delay time;
A first echo amount measuring step for measuring an echo amount of the first audio signal;
A first distortion amount measuring step of measuring a distortion amount of the first audio signal;
Have
The second quality deterioration index calculating device step includes:
A second signal generating step for generating a second audio signal;
A second echo amount measuring step for measuring an echo amount of the second audio signal;
A second distortion amount measuring step for measuring a distortion amount of the second audio signal;
Have
The transmission psychological deterioration amount calculating step calculates a first transmission psychological deterioration amount from the delay time and an echo amount of the first audio signal based on the transmission sound rate, and the audible sound Based on the rate, the second transmission psychological deterioration amount is calculated from the delay time and the echo amount of the second audio signal,
The listening psychological deterioration amount calculating step calculates a first listening psychological deterioration amount from a distortion amount of the first sound signal based on the listening sound rate, and calculates a second sound based on the transmission sound rate. Calculate the second listening psychological degradation amount from the distortion amount of the signal,
The integrated psychological deterioration amount calculating step includes the first transmission psychological deterioration amount, the second transmission psychological deterioration amount, the first listening psychological deterioration amount, the second listening psychological deterioration amount, and the delayed psychology. The call quality estimation method according to claim 3, wherein an integrated psychological deterioration amount is calculated from the deterioration amount. The transmission sound rate measurement step measures the transmission sound rate by dividing the transmission signal every unit time and measuring the rate at which the volume of the transmission signal per unit time exceeds a threshold value. ,
The audible sound rate measuring step measures the audible sound rate by dividing the audible signal every unit time and measuring a ratio at which the volume of the audible signal per unit time exceeds a threshold value. 5. The call quality estimation method according to any one of items 1 to 4.   The speech switching frequency measurement step divides the transmission signal and the listening signal for each unit time, and the number of times the volume of the transmission signal per unit time continuously exceeds the threshold and the volume of the listening signal per unit time are continuous. The call quality estimation method according to claim 1, wherein the number of utterance switching is measured based on the number of times exceeding a threshold. The transmission psychological deterioration amount calculating step calculates a transmission psychological deterioration amount using weighting based on the transmission sound rate,
The listening psychological deterioration amount calculating step calculates a listening psychological deterioration amount using weighting based on the listening sound rate,
The call quality estimation method according to any one of claims 1 to 6, wherein the delayed psychological deterioration amount calculating step calculates a delayed psychological deterioration amount using weighting based on the number of times of utterance switching. A call quality estimation device for estimating quality in an evaluation target system of a voice call service,
A transmission sound rate measurement unit for measuring one transmission sound rate of the evaluation target system;
A listening sound rate measuring unit for measuring one listening sound rate of the evaluation target system;
An utterance switching number measuring unit for measuring the number of utterance switching per unit time between transmission and listening in the evaluation target system;
A transmission psychological deterioration amount calculation unit that calculates a transmission psychological deterioration amount from a quality deterioration index based on the transmission sound rate;
A listening psychological deterioration amount calculating unit for calculating a listening psychological deterioration amount from a quality deterioration index based on the listening sound rate;
A delayed psychological deterioration amount calculating unit for calculating a delayed psychological deterioration amount from a quality deterioration index based on the number of utterance switching times;
An integrated psychological deterioration amount calculating unit that calculates an integrated psychological deterioration amount from the transmitted psychological deterioration amount, the listening psychological deterioration amount, and the delayed psychological deterioration amount;
A subjective quality evaluation value calculation unit for converting the integrated psychological deterioration amount into a subjective quality evaluation value;
A call quality estimation device comprising: A call quality estimation system for estimating the quality of a voice call service evaluation target system,
A first quality deterioration index calculating device connected to one of the evaluation target systems and measuring one quality deterioration index of the evaluation target system;
A second quality deterioration index calculating device connected to the other of the evaluation target system and measuring the other quality deterioration index of the evaluation target system;
A subjective quality evaluation value calculating device connected to the first quality deterioration index calculating device and the second quality deterioration index calculating device;
Have
Subjective quality evaluation value calculation device
A transmission sound rate measurement unit for measuring one transmission sound rate of the evaluation target system;
A listening sound rate measuring unit for measuring one listening sound rate of the evaluation target system;
An utterance switching number measuring unit for measuring the number of utterance switching per unit time between transmission and listening in the evaluation target system;
Based on the transmission sound rate and the listening sound rate, a transmission psychological deterioration amount calculation unit that calculates a transmission psychological deterioration amount from a measured quality deterioration index;
Based on the transmission sound rate and the listening sound rate, a listening psychological deterioration amount calculating unit that calculates a listening psychological deterioration amount from a measured quality deterioration index;
A delayed psychological deterioration amount calculation unit that calculates a delayed psychological deterioration amount from the measured quality deterioration index based on the number of utterance switching times;
An integrated psychological deterioration amount calculating unit that calculates an integrated psychological deterioration amount from the transmitted psychological deterioration amount, the listening psychological deterioration amount, and the delayed psychological deterioration amount;
A subjective quality evaluation value calculation unit for converting the integrated psychological deterioration amount into a subjective quality evaluation value;
A call quality estimation system. The first quality deterioration index calculating device includes:
A first signal generator for generating a first audio signal;
A delay time measurement unit for measuring the delay time;
A first echo amount measurement unit for measuring an echo amount of the first audio signal;
A first distortion amount measuring unit for measuring a distortion amount of the first audio signal;
Have
The second quality deterioration index calculation device includes:
A second signal generator for generating a second audio signal;
A second echo amount measurement unit for measuring an echo amount of the second audio signal;
A second distortion amount measurement unit for measuring the distortion amount of the second audio signal;
Have
The transmission psychological deterioration amount calculation unit calculates a first transmission psychological deterioration amount from the delay time and the echo amount of the first audio signal based on the transmission sound rate, and the audible sound Based on the rate, the second transmission psychological deterioration amount is calculated from the delay time and the echo amount of the second audio signal,
The listening psychological deterioration amount calculation unit calculates a first listening psychological deterioration amount from a distortion amount of the first sound signal based on the listening sound rate, and a second sound based on the transmission sound rate. Calculate the second listening psychological degradation amount from the distortion amount of the signal,
The integrated psychological deterioration amount calculation unit includes the first transmission psychological deterioration amount, the second transmission psychological deterioration amount, the first listening psychological deterioration amount, the second listening psychological deterioration amount, and the delayed psychology. The call quality estimation system according to claim 9, wherein an integrated psychological deterioration amount is calculated from the deterioration amount.

Images