JPH0786993A - Echo calceler - Google Patents

Abstract

PURPOSE:To reduce power consumption by remarkably reducing processing computing quantity relating to the update of a tap coefficient and to improve echo suppression characteristic by improving computing accuracy in filter processing. CONSTITUTION:True tap data and tap coefficient for a response part are selected from the impulse response of an echo path, and also, selected tap data and tap coefficient are used. In this way, a false echo can be generated by applying the filter processing to a false echo generation means 3, while, a control means 6 which performs control so as to update only the selected tap coefficient by a tap coefficient update means 5 is provided.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an echo canceller suitable for removing an echo that occurs when a telephone call or the like is made and causes deterioration of the call quality.

[0002]

2. Description of the Related Art Conventionally, when a telephone call or the like is performed, an echo is generated due to the contact between the handset and the two-line / four-line conversion of the transmission line, which causes the deterioration of the call quality.
To prevent this, an echo canceller for suppressing echo has been developed.

This echo canceller suppresses an echo when it is anxious due to a long transmission line and a large transmission delay, such as a two-way repeater for a call transfer service or a repeater for a telephone conference. Is effective when used,
Widely used. Further, in recent years, applications in which band compression is performed by a voice CODEC in order to reduce the transmission capacity have been widely put into practical use.
It is also widely applied to eliminate echo caused by the delay of DEC. In practice, it has come to be widely used in wireless networks such as mobile terminals and their base station systems.

FIG. 4 is a diagram showing the structure of a general echo canceller which has been conventionally used.
In, since the reception side signal enters the transmission side signal as an echo signal, if the portion that outputs this echo signal to the transmission side signal is virtually the echo generation section 36, an echo component is generated in this echo generation section 36, In the echo component adding section 37, this echo component is added to the audio signal (transmission signal) from the partner terminal and output.

The echo canceller 30 receives the input signal on the receiving side and predicts the echo path characteristic of the echo generating section 36 based on the input signal on the receiving side using an FIR type adaptive filter. Remove the ingredients,
This is output as a transmission side output signal, and for this reason, the tap coefficient storage unit 31 and the tap storage unit 3 are provided.
2, a pseudo echo generating unit 33, an echo component removing unit 34, and a tap coefficient updating unit 35.

Here, the tap coefficient storage unit 31 stores tap coefficients for performing filtering processing on the reception side input signal, and the tap storage unit 32 stores tap data for the reception side input signal. It is for temporary storage. In addition, the pseudo echo generation unit 33 includes the tap coefficient stored in the tap coefficient storage unit 31 and the tap storage unit 3.
By using the tap data temporarily stored in 2, the filter processing is performed and a pseudo echo is generated.

The echo component removing section 34 obtains the transmission side output signal from which the echo component is removed by obtaining the difference between the pseudo echoes generated by the pseudo echo generating section 33 for the transmission side input signal containing the echo component. It is supposed to be done. Further, the tap coefficient updating unit 35 updates the tap coefficient for performing the filtering process on the next reception side input signal by using the transmission side output signal output from the subtraction unit 34, and specifically, Has been updated using equation (1).

H_i ^{(n + 1)}= H_i ⁽ⁿ⁾+ {(Α ・ e_n) / Σ (X_ni)²} X_ni ... (1) where H_iIs tap coefficient, i is tap number, X_nIs received
Signal, n is time, e _nIs the residual echo signal, α is the updater
Is a number. With this configuration, conventional echo cancel
In the Sera, for example, impulse response as shown in FIG.
For echo paths that have answers, remove the echo component
There is.

As shown in FIG. 5, the impulse response of the echo path consists of a simple delay portion (a) and a response portion (b) following it. The delay part (a) corresponds to the signal delay depending on the distance between the echo generation point and the echo canceller, but the required filter tap number is determined with respect to the impulse response time of the echo path to be suppressed. To be done.

By the way, in the case of an echo generated by 2-line to 4-line conversion, the response section is actually 1 as shown in FIG.
It is set to 0 msec to 50 msec. Also, when inserting a speech encoder having a high degree of compression, a delay of up to about 50 msec may occur. Therefore, the number of taps of the echo canceller actually used corresponds to this,
It is often around 400 taps corresponding to 40 to 50 msec.

[0011]

However, in such a conventional echo canceller, since all the tap coefficients including the entire response portion are filtered, all the tap coefficients must be updated. However, a large amount of calculation is required, and the total amount of calculation of the echo canceller may be enormous.

For this reason, the required memory amount and the required processing calculation amount become enormous, and the miniaturization is indispensable not only when it is applied to a repeater, but also when it is applied to a loudspeaker telephone, especially a loudspeaker car telephone. ， There is a problem that is an obstacle to low power consumption. Further, the tap coefficient held by the echo canceller and the mere delay portion (see (a) in FIG. 5) of the taps and the excess taps after the response portion cause a decrease in the convergence speed and a deterioration in the echo suppression amount. There is also a problem that it may occur.

The present invention was devised in view of the above problems, and significantly reduces the amount of processing calculation related to the update of the tap coefficient to reduce power consumption and improve the calculation accuracy in filter processing. It is an object of the present invention to provide an echo canceller with improved echo suppression characteristics.

[0014]

FIG. 1 is a block diagram of the principle of the present invention. In FIG. 1, reference numeral 1 is a tap coefficient storage means, and this tap coefficient storage means 1 stores tap coefficients. is there. Further, reference numeral 2 denotes a tap storage means, and the tap storage means 2 temporarily stores tap data for a receiving side input voice signal.

Further, 3 is a pseudo echo generating means,
The pseudo echo generating means 3 uses the tap data temporarily stored in the tap storing means 2 and the tap coefficient stored in the tap coefficient storing means 1 to perform a filtering process to generate a pseudo echo. It is what makes me. Reference numeral 4 denotes an echo component removing means, which uses the pseudo echo generated by the pseudo echo generating means 3 to remove the echo component contained in the transmission side input audio signal. is there.

Further, 5 is a tap coefficient updating means,
The tap coefficient updating means 5 updates the tap coefficient for performing the filtering process on the next reception side input audio signal by using the audio signal output from the echo component removing means 4. Further, 6 is a control means, and this control means 6 selects and selects tap data and a tap coefficient corresponding to a true response section from the impulse response of the transmission side input audio signal including the echo component. By using the tap data and the tap coefficient, the pseudo echo generating means 3 performs the filtering process to generate the pseudo echo, while the tap coefficient updating means 5 updates only the selected tap coefficient. It is controlled (claim 1).

Further, in the echo canceller of the present invention, the control means 6 determines the start timing identification means for identifying the input timing of the impulse response of the input voice signal on the transmission side including the echo component, and the tap coefficient from the impulse response. Using the selecting means for selecting, the partial pseudo echo generating means for generating a partial pseudo echo by using the selected tap coefficient, and the pseudo echo generated by the partial pseudo echo generating means, It comprises a partial echo component removing means for removing an echo component contained in the input audio signal and an echo suppression amount determining means for determining whether or not the echo component is effectively removed by the partial echo component removing means. (Claim 2).

Further, in the echo canceller of the present invention, the tap coefficient stored in the tap coefficient storage means 1 is
It is also possible to divide the set into a plurality of sets in a time series and control the control means 6 to select one of the plurality of sets (claim 3).

[0019]

In the above echo canceller of the present invention, the tap coefficient storage means 1 stores the tap coefficient, and the tap storage means 2 temporarily stores the tap data for the receiving side input voice signal. Further, in the pseudo echo generating means 3, the tap data temporarily stored in the tap storing means 2 and the tap coefficient stored in the tap coefficient storing means 1 are used to perform a filtering process to generate a pseudo echo. It is what is generated.

The echo component removing means 4 uses the pseudo echo generated by the pseudo echo generating means 3 to remove the echo component contained in the transmitting side input audio signal. Further, the tap coefficient updating means 5 updates the tap coefficient for performing the filtering process on the next input audio signal using the audio signal output from the echo component removing means 4.

Further, the control means 6 selects the tap data and the tap coefficient corresponding to the true response section, and
By using the selected tap data and tap coefficient, the pseudo echo generating means 3 performs the filtering process to generate the pseudo echo, while the tap coefficient updating means 5 updates only the selected tap coefficient. (Claim 1).

Further, in the echo canceller of the present invention, the determination start time identifying means identifies the input timing of the impulse response of the transmitting side input voice signal containing the echo component, and the selecting means selects the impulse response and the echo described later. A tap coefficient is selected from the determination result from the suppression amount determination means. In the partial pseudo echo generating means, a partial pseudo echo is generated using the selected tap coefficient,
The partial echo component removing means removes the echo component contained in the transmitting side input audio signal by using the pseudo echo generated by the partial pseudo echo generating means.

The echo suppression amount determining means determines whether or not the echo component is effectively removed by the partial echo component removing means, and the determination result is output to the selecting means.
(Claim 2). Furthermore, the echo canceller of the present invention is
The tap coefficient stored in the tap coefficient storage means 1 is divided into a plurality of sets in time series, and the control means 6 controls to select one of the plurality of divided sets (claim) Item 3).

[0024]

FIG. 2 is a block diagram showing an embodiment of the present invention. In FIG. 2, 22 is an echo generator,
Reference numeral 23 is an echo component addition unit. That is, the echo component generated due to the sound contact between the handset and the two-wire / four-wire conversion enters the transmission side signal as the echo signal from the reception side signal as described above. The portion that outputs the signal is virtually defined as the echo generator 22.

Therefore, the echo component is generated in the echo generating unit 22, and is included in the voice signal (transmission signal) from the partner terminal and output in the echo component adding unit 23. The echo generator 22
The echo component adding section 23 is added to explain the mechanism of echo generation, and is different from the constituent elements of the present invention.

Numeral 10 is an echo canceller installed in the exchange. This echo canceller is for removing (suppressing) the echo component contained in the transmission signal from the echo component adding section 23, and is a tap. Coefficient storage unit (tap coefficient storage unit) 11, tap storage unit (tap storage unit) 12, pseudo echo generation unit (pseudo echo generation unit) 13, and echo component removal unit (echo component removal unit) 1
4 and a tap coefficient updating unit 15 and a control unit 16.

Here, the tap coefficient storage unit 11 stores the tap coefficient, and the tap storage unit 12 is also provided.
Receives a received signal and stores the received signal as tap data (tap). Further, the pseudo echo generating unit 13 performs filtering processing using the tap coefficients stored in the tap coefficient storage unit 11 and the taps stored in the tap storage unit 12 to include in the transmission signal. The predicted echo component is predicted and the predicted echo component is output as a pseudo echo.

Further, the echo component removing section 14 receives the transmission signal containing the echo component and the pseudo echo from the pseudo echo generating section 13 and obtains the difference between the transmission signal and the pseudo echo. The transmission signal from which the echo component is removed can be output. Further, the tap coefficient updating unit 15 updates the tap coefficient stored in the tap coefficient storage unit 11 using the transmission signal from the echo component removing unit 14, and specifically, in the above-described conventional case. Similar to the above, it has been updated using equation (1). As a result, the echo component of the received signal input next time can be predicted.

Further, the control section 16 selects tap data and tap coefficient corresponding to the true response section from the impulse response of the echo path (impulse response of the input voice signal on the transmission side containing the echo component), and By using the selected tap data and the tap coefficient, the pseudo echo generating unit 13 performs the filtering process to generate the pseudo echo, while the tap coefficient updating unit 15 updates only the selected tap coefficient. To control it. In other words, by inputting a transmission signal containing an echo component and using a small number of taps and tap coefficients, a tap and a tap coefficient that can most effectively remove the echo component are selected, and the pseudo echo generating means 13 It controls to generate a pseudo echo.

Functionally, the control unit 16 is functionally a determination start time identification unit (determination start time identification means) 17, a determination target selection unit (selection unit) 18, a partial pseudo echo generation unit (partial pseudo). It is divided into an echo generation means) 19, a subtraction section (partial echo component removal means) 20, and an echo suppression amount determination section (echo suppression amount determination section) 21. Here, the determination start time point identifying unit 17 determines the transmission signal including the echo component,
The reception signal is input, and the input timing of the impulse response of the transmission signal input by the determination target selection unit 18, which will be described later, is identified.

The determination target selection unit 18 selects from the tap coefficient storage unit 11 a tap coefficient that can most effectively generate a pseudo echo capable of removing an echo component based on the impulse response of the transmission signal. . The determination target selection unit 18 divides the tap coefficients stored in the tap coefficient storage unit 11 into four sets (blocks) in chronological order as illustrated in FIG. 3, and the divided blocks are divided into four sets (blocks). It is possible to select any one of them, which makes it possible to eliminate the processing relating to the simple delay portion as shown in FIG.

For example, as shown in FIG. 3, tap coefficient cases
Four tap coefficients stored in the storage unit 11 in chronological order
When divided into blocks, the first block is chronologically ordered.
It is divided like a block 41 to a fourth block 44. Also,
The tap coefficient is H and the number of taps per unit block is
If n, the first block is H₀~ H_n-1, Second block
Ku is H_n~ H_2n-1, The third block is H_2n~ H_3n-1, No.
4 blocks are H_3n~ H _4n-1Becomes

In the above case, the determination target selection unit 18
The selection mode of the block by (a) to
There are 10 possible ways of (j). (A) When selecting the first block 41. (B) When selecting the second block 42. (C) When selecting the third block 43.

(D) When selecting the fourth block 44. (E) When selecting the first block 41 and the second block 42. (F) When selecting the second block 42 and the third block 43. (G) When selecting the third block 43 and the block 4. (H) When selecting the first block 41, the second block 42, and the third block 43.

(I) Second block 42 and third block 4
When selecting 3 and the fourth block 44. (J) When selecting the first block 41, the second block 42, the third block 43, and the fourth block 44. By the way, the partial pseudo echo generation unit 19 performs a filter process using the tap coefficients in the block selected by the determination target selection unit 18 and the taps corresponding to these tap coefficients to generate a partial pseudo echo. It is what is generated.

The subtracting section 20 includes a partial pseudo echo generating section 1
Partial pseudo echo from 9 and echo component addition unit 23
The transmission signal including the echo component from the input signal is input, and the difference between these signals is obtained, so that the signal is output to the echo suppression amount determination unit 21 as a signal from which the echo component corresponding to the selected block is removed. Has become. The transmission signal obtained by subtracting the echo components from the subtraction unit 20 is generated for each tap coefficient according to the selection mode of the 10 types of blocks selected by the determination target selection unit 18, and the echo suppression amount determination is performed. The data is output to the section 21.

Then, the echo suppression amount 21 is calculated by the subtraction unit 20.
The signals from which the 10 echo components have been removed are input, the signal having the best echo suppression characteristic is determined, and the determination result is output to the determination target selection unit 18. Here, when there are a plurality of selection modes of blocks that can effectively remove the echo component, the one having a small number of selected blocks is adopted.

With the above configuration, the echo canceller according to the embodiment of the present invention operates as follows.
That is, when a call is connected between the transmission side terminal and the reception side terminal (not shown), the reception signal is input to the echo generation unit 22 from the reception side terminal. When the transmission signal is output from the transmission-side terminal, the transmission signal includes the echo component generated in the echo generation unit 22 and is input to the echo canceller 10.

At this time, in the echo canceller 10, the determination start time identification unit 17 receives the transmission signal, and the control signal to start the estimation of the echo path is output to the determination target selection unit 18. This is because the transfer characteristic of the echo path is given by the line to which the echo canceller 10 is connected. The identification of the call start time by the determination start time identification unit 17 can identify the time when the call is disconnected and the time when the signal is first input to the receiving side after that as the call start time.

By the way, in the echo canceller 10, when the above-mentioned received signal is input, information about this received signal is stored in the tap storage unit 12 as tap data. In the tap coefficient storage unit 11, the tap coefficients stored in advance are stored in four blocks 4
It is divided into 1 to 44. Next, for the time t from the start of the call, the echo path is estimated and the echo is estimated by using all the tap coefficients stored in advance in the tap coefficient storage unit 11 and all the tap data stored in the tap storage unit 12. Suppress the.

That is, in the pseudo echo generator 13, a filter process is performed using all tap coefficients and tap data to generate a pseudo echo. Then, the echo component removing unit 14 suppresses the echo component of the transmission signal by obtaining the difference between the transmission signal including the echo component and the pseudo echo. At the time r after the time t has elapsed from the start of communication, the echo path is estimated and the echo is suppressed by all taps, and the delay among the stored tap coefficients is, for example, as shown in (a) of FIG. Except for the part and the like, the part corresponding to the true response part as shown in FIG.

Specifically, using the blocks divided and stored in the tab coefficient storage unit 11 into four,
The determination target selection unit 18 inputs the tap coefficients and the tap data corresponding to the combinations of the 10 blocks from (a) to (j). Then, the partial pseudo echo generating unit 19 generates a partial pseudo echo using the selected tap coefficient and tap data. Then, the subtraction unit 20 obtains the difference between the transmission signal including the echo component and the partial pseudo echo, and outputs the transmission signal to the echo suppression amount determination unit 21 as the echo component being suppressed.

In the echo suppression amount judging section 21, the above-mentioned 10
Corresponding to the combinations of the blocks, the combinations of the blocks having the best echo suppression characteristics are selected from the signals input from the subtraction unit 20. At this time, if the echo suppression characteristics are almost the same, a block having a small number of blocks is selected. This makes it possible to select taps and tap coefficients corresponding to only the true response portion as shown in FIG. 5B, except for the delay portion as shown in FIG.

In the processing within this r time, since the pseudo echo calculation when echo suppression is performed at all taps is continuously performed during the time t after the call is connected, the control unit 16 Only the processing for calculating the residual echo and the suppression amount corresponding to each block according to (4) is added within r time. When the call as described above is connected and the processing at the time t + r is completed and the tap coefficient for the true response section is identified, the pseudo echo generating section 13 generates a pseudo echo using only the tap coefficient and the corresponding tap. generate. After that, the echo component removing unit 14 removes the echo component as in the case described above, and outputs this to the receiving side as a transmission signal.

Further, the tap coefficient updating unit 15 updates only the tap coefficient identified as the tap coefficient for the response portion truly necessary for echo suppression.
The subsequent calculation of the pseudo echo is also performed using only the corresponding tap coefficient and the corresponding tap until the call is disconnected. Therefore, tap coefficients that do not correspond stop the updating operation thereafter and are not used for pseudo echo calculation.

As described above, according to the echo canceller according to the embodiment of the present invention, after t + r time has elapsed since the call was connected, the pseudo echo generating section 13 has a portion other than the true response portion such as a delay portion. It is not necessary to carry out the filtering process related to. Therefore, it is possible to significantly reduce the amount of processing calculation and power consumption, and at the same time, prevent the characteristic deterioration caused by the calculation error due to unnecessary taps after the delay portion and the true response portion, and improve the echo suppression characteristic. There is an advantage that can be achieved.

[0047]

As described in detail above, the first aspect of the present invention
According to the echo canceller described above, tap coefficient storage means for storing tap coefficients, tap storage means for temporarily storing tap data for an input audio signal, tap data temporarily stored in the tap storage means, and By using the tap coefficient stored in the tap coefficient storage means, by performing a filtering process, using the pseudo echo generating means for generating a pseudo echo, and the pseudo echo generated by the pseudo echo generating means, Echo component removing means for removing the echo component contained in the input audio signal on the transmission side, and the tap coefficient for performing the filtering process for the next input audio signal are updated using the audio signal output from the echo component removing means. It is configured to include tap coefficient updating means that outputs the output audio signal on the transmitting side from which the echo component is removed. In the echo canceller configured as described above, the tap data and the tap coefficient corresponding to the true response portion are selected from the impulse response of the input voice signal on the transmission side including the echo component, and the selected tap data and tap coefficient are selected. By using and, the pseudo echo generating means performs a filter process to generate a pseudo echo, and a control means for controlling the tap coefficient updating means to update only the selected tap coefficient. As a result, it is possible to significantly reduce the amount of processing calculation and power consumption, and at the same time, prevent the characteristic deterioration caused by the calculation error due to unnecessary taps after the delay part and the true response part, and suppress the echo suppression characteristics. There is an advantage that can be improved.

Further, according to the echo canceller of the second aspect of the present invention, the control means includes a determination start time point identification means for identifying the input timing of the impulse response of the transmission side input audio signal including the echo component, Selecting means for selecting a tap coefficient from the impulse response; partial pseudo echo generating means for generating a partial pseudo echo using the selected tap coefficient; and a partial pseudo echo generating means for generating the partial pseudo echo. Using the pseudo echo, a partial echo component removal means for removing the echo component contained in the input audio signal on the transmission side and a judgment as to whether or not the echo component is effectively removed by the partial echo component removal means Since it is configured with an echo suppression amount determination means, a simple device design allows calculation by unnecessary taps after the delay part and the true response part. Prevent characteristic deterioration resulting from the difference, an improvement in echo suppression characteristics, effectively can be advantageously realized.

Further, according to the echo canceller of the third aspect of the present invention, the tap coefficient stored in the tap coefficient storage means is divided into a plurality of sets in chronological order.
In the control means, by controlling so as to select any one of the plurality of divided sets, the true response part can be identified by a simple process, so that the echo suppression characteristic can be further improved. There is an advantage that can be achieved.

[Brief description of drawings]

FIG. 1 is a principle block diagram of the present invention.

FIG. 2 is a block diagram showing an embodiment of the present invention.

FIG. 3 is a diagram illustrating a tap coefficient selection procedure of a determination target selection unit according to an embodiment of the present invention.

FIG. 4 is a block diagram showing a conventional example.

FIG. 5 is a diagram showing an impulse response example of an echo path.

FIG. 6 is a diagram showing an impulse response example of an echo path.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 tap coefficient storage means 2 tap storage means 3 pseudo echo generation means 4 echo component removal means 5 tap coefficient update means 6 control means 10 echo canceller 11 tap coefficient storage section 12 tap storage section 13 pseudo echo generation section 14 subtraction section 15 tap coefficient Update unit 16 Control unit 17 Determination start point identification unit 18 Determination target selection unit 19 Partial pseudo echo generation unit 20 Subtraction unit 21 Echo suppression amount determination unit 22 Echo generation unit 23 Echo component addition unit 30 Echo canceller 31 Tap coefficient storage unit 32 Tap storage unit 33 Pseudo echo generation unit 34 Subtraction unit 35 Tap coefficient storage unit 36 Echo generation unit 37 Echo component addition unit 41 to 44 blocks

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Toshinobu Takashima 1015 Kamiodanaka, Nakahara-ku, Kawasaki-shi, Kanagawa, Fujitsu Limited (72) Inventor Fumiaki Nishida 1015 Kamedotachu, Nakahara-ku, Kawasaki, Kanagawa Prefecture, Fujitsu Limited (72) Inventor Rika Nishiike 1015 Kamiodanaka, Nakahara-ku, Kawasaki City, Kanagawa Prefecture Fujitsu Limited

Claims (3)

[Claims] 1. A tap coefficient storage means (1) for storing tap coefficients, a tap storage means (2) for temporarily storing tap data of an input audio signal, and a temporary storage in the tap storage means (2). Pseudo echo generating means (3) for performing a filtering process to generate a pseudo echo by using the tap data and the tap coefficient stored in the tap coefficient storing means (1).
And an echo component removing means (4) for removing an echo component included in the output audio signal by using the pseudo echo generated by the pseudo echo generating means (3), and a filtering process for the next input audio signal. The tap coefficient to be applied is configured to include tap coefficient updating means (5) for updating the tap coefficient using the audio signal output from the echo component removing means (4), so that the echo component removed output is obtained. In an echo canceller capable of outputting a voice signal, selecting tap data and a tap coefficient corresponding to a true response part from an impulse response of an echo path, and using the selected tap data and tap coefficient The pseudo tap generating means (3) applies a filtering process to generate a pseudo echo, while the selected tap group is selected. The is characterized in that includes a control for controlling means (6) so as to update at the tap coefficient updating means (5) only, the echo canceller. 2. The control means (6) discriminating start time point discriminating means for discriminating the input timing of the impulse response of the echo path, selecting means for selecting a tap coefficient from the impulse response, and the selected tap coefficient. By using the partial pseudo echo generating means for generating a partial pseudo echo, and a portion for removing an echo component contained in the output audio signal by using the pseudo echo generated by the partial pseudo echo generating means. 2. An echo echo component removal means and an echo suppression amount determination means for determining whether or not the echo component is effectively removed by the partial echo component removal means. Echo canceller. 3. The tap coefficient stored in the tap coefficient storage means (1) is divided into a plurality of sets in time series order, and the control means (6) divides the plurality of divided sets. The echo canceller according to claim 1 or 2, wherein the echo canceller is controlled so as to select either one.