No relevant resource is found in the selected language.

This site uses cookies. By continuing to browse the site you are agreeing to our use of cookies. Read our privacy policy>Search


To have a better experience, please upgrade your IE browser.


There is echo when in a point-to-point call

Publication Date:  2016-07-31 Views:  2651 Downloads:  0

Issue Description

Customer feedback that when deliver a project, There is echo when in a point-to-point call.

Alarm Information

no alarm

Handling Process

First make clear the customer`s live environment.
1. Customer live environment they are using 2 TE40 to call each other. The version is V5R2SPC200.
2. Customer is using 3rd party TV of LG 55UF680V, which is non-standard RP solution.
3. Customer is using 2 M220 cascaded for audio input.
4. Customer is using embeded speaker(HDMI output), there is no other external speaker.
5. Customer`s operation is very easy, just call remote side, after connected, local side can hear the echo.
6. Customer use Cisco terminal, there is little echo which is better than huawei TE40.
For the echo problem, we know that
If local side hear the echo, except the local audio loop back(low possibility, we don`t discuss a lot in this case), local sound go through the remote speaker output, then go into the remote MIC, then it go back local site again. Which caused that local site can hear the local sound.

    In fact there is a easy way to know the root cause, we can move the MIC in remote side a little far away from speaker, even mute the MIC, the echo will be decreased even disappear. So we can confirm whether the echo is coming from the remote side.
     This case we analyse below:
When in a conference, the Sound of Room2(we will use S2 for short, below is same) is collected by the MIC1 in the Room1, and then send back to the speaker in Room2. So Room2 can hear his own sound. It is caused by the multipath such as reflect, diffract etc. You can find the picture as below.

When the S2 send to the speaker1, in room1 it will also send to EC(echo canceller) as a Room 2 reference signal. And the MIC1 in the room1 will collect S1, which is not only the Room1 sound but also the Room2 echo. And S1 will be handled by EC(echo canceller) and then will have a S1’ output after echo cancellation. Then to the Room2.
echo based on the relationship with the speaker sound and the mutilpath echo coming from the speaker sound. First codec will build the model of S2. Then use it to estimate the echo, in the end minus this estimated value from S1.

We have the following requirement for the echo sound and reference signal.
1. The echo sound and reference signal can not be distorted. Distortion include the following possiblities, one is the echo sound and reference signal can not be full-scale. The other is that the echo sound can not be Sound Processed. Or the frequence of echo sound will be changed, so it can not offset with the reference signal.
2. EC(echo canceller) will cache the reference signal, then will be offset after the echo sound comes. Now we require this time of cache no more than 100ms.
3. reverberation of echo sound is no more than 250ms, if the reverberation is a little big, caused that echo sound is 1000ms, after the reference signal is finished, the echo is still persists, so the echo will be heard.
Based on the above information, we will collect the information(such as the verion  V500R002C00SPC303T)
Let`s use a point-to-point call. A site and B site (please only use 1 M220 in site X). When A site says words, A site can hear his own words. Now we need to collect the info from site B.
1.SSH into site B, username is debug and password is debug or Change_Me by default.
Execute audiotest sendTestData on 9000 102 680 3 is the address of your PC which installs AudioCapTool.exe this software.

2.Run AudioCapTool.exe in your PC(,
Remote IP:   this IP is TE40 IP.
Port 9000  by default.
Click `listen`. When you start capture, it will prompt in below red line. Meantime, in the same path of AudioCapTool.exe, it will generate the file sin.pcm and rin.pcm.

3.When you collect the info, A site should keep talking. After says one sentence, he should confirm whether there is echo or not. If there is  echo, please say `echo`, If there is no, please say `no echo`.
After 20 seconds, then clap, after 5 seconds, clap again, clap for 10 times.
4.Close the AudioCapTool.exe, execute the following commands and save the traces file.
mc audio maec get
mnt maecThrd get
trace setdebugcpuid 2

audio getmaecinfo                         //execute 5 times
audio getqueinfo 122                     //execute 3 times
audio getqueinfo                            //execute 3 times
audio getcoefInfo                           // execute 3 times
audio getaitaskinfo
audio getmaecstatus
audio getmaecstatus 1
audio getmaecstatus 2
audio getmaecstatus 3
audio getmaecstatus 4

Root Cause

from the traces and packets, we found that
we compared the reference signal and echo signal, the echo signal is 260ms delayed than reference signal. As you can see following picture, what TE40 requires of cash time is within 100ms, so the echo can not disappear.
We can try to change echo delay settings to avoid this problems.
System settings---input output---audio---echo delay

Environment Reverberation is too big. While huawei TE40 can only support within 250ms, we checked the traces customer feedback, Reverberation of some sites are within 200ms, while some sites are reached into 1000ms, it`s already exceed the maximum of EC(echo canceller).
For this we have to
a. Release a new version to adapt the environment.
b. Environment rectification


Because customer`s tv is 3rd party TV, it`s not standard RP, so we suggest customer configure the echo delay into 5. After configure it, the echo issue is optimized a lot. In addition, we feedback this issue into R&D, and send the test version (V500R002C00SPC303T)which optimize echo issue to customer, customer test it ok. we will release a commercial version TEX0 V500R02C00SPC600 in the end of August.


Be familiar with the normal diagnose way.