openmeetings-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Irina Arkhipets <ia...@unipro.ru>
Subject Re: AEC in the new release
Date Mon, 29 Apr 2013 14:01:07 GMT
Hi All,

First of all, thank you for your advices and sorry for so long report :)

We have made a first test with the different parameters.

I would like to share the results.

Briefly:

  *

    Sound quality with echoPath=0 is generally better than for
    echoPath=128 and 256 (we lost sound of one attendee sometime)

  *

    Echo is actually suppressed with echoPath 128/256 whereas it is
    noticeable with echoPath = 0 on the same configuration (using
    notebook on the client side)

  *

    Delay between the video and sound is less for echoPath = 0

  *

    Delay between the video and sound is less for H263

So the conclusions are:

  *

    For the conference where some users with notebooks echoPath = 128 is
    more preferrable;

  *

    For the conference where notebooks are not used echoPath = 0 is more
    preferrable;

  *

    For the conferences with many attendees H263 is more preferrable
    because it seems like that H264 usage increases the delay.

*Detailed report about the testing:*

Date: 29-April-2013

Where: demo.dataved.ru

Participants:

  *

    Artyom Horuzhenko (Linux desktop)

  *

    Denis Kandrov (Linux Desktop)

  *

    Irina Arkhipets (Windows desktop)

  *

    Yulia Algaer (Windows notebook)

  *

    SIP integration was turned on in the room, but nobody called there
    during the testing. SIP transport presented in the user list.

Tested parameters: echoSuppression (true/false), echoPath (0/128/256), 
codecType (264/263)

_*Results:*_

*Test 1.* echoPath=128, echoSuppression=true, video codec=264

(These are default values in the system)

  *

    No echo during all the test;

  *

    Sound from Denis disappeared sometime;

  *

    Delay between the hand clapping and its sound is less than 1 second,
    but noticeable.

*Test 2.* echoPath=0, echoSuppression=true, video codec=264

  *

    Echo after Yulia (user with notebook) entered the room

  *

    Sound from Denis did not disappear all the time

  *

    No visible delay between the hand clapping and its sound

*Test 3.* echoPath=256, echoSuppression=true, video codec=264

  *

    No echo during all the test

  *

    Sound from Denis did not disappear all the time

  *

    Delay between the hand clapping and its sound is less than 1 second,
    but noticable

  *

    Sound quality is significantly worse than in the Test 1.

  *

    SIP-transport produced noticeable noise

*Test 4.* echoPath=0, echoSuppression=false, video codec=264

  *

    Echo after Yulia (user with notebook) entered the room

  *

    Strong "jungle" was hearable all the time, sound quality is
    significantly worse than in all the other tests.

  *

    Sound from Denis did not disappear all the time

*Test 5.* echoPath=128, echoSuppression=false, video codec=264

  *

    No echo

  *

    When Yulia (user with notebook) spoke, some "jungle" was hearable;

  *

    Delay between the hand clapping and its sound is less than 1 second,
    but noticeable

*Test 6.* echoPath=128, echoSuppression=true, video codec=263

  *

    No echo during all the test

  *

    Delay between the hand clapping and its sound is almost negligible

  *

    If I increased video size, quality is visible worse than with H264
    under the same conditions

Best regards,

Irina.


On 26.04.2013 18:04, Irina Arkhipets wrote:
> Hi All,
>
> Artyom just made some changes and for now AEC parameters can be
> configurable via config.xml.
>
> So we are planning to have some testing on demo.dataved.ru on the nearest
> Monday (April, 29).
>
> Approximate time is 15:00 NSK (GMT +7), and still we can move the time if
> it's inconvenient.
>
> If somebody from the list would like to participate - you are welcome, just
> let me know :).
>
> The testing scenario would be to enter the room and do the same things with
> the different settings:
>
>     -
>
>     No echo cancellation (echoPath=0) and all the other settings are default
>     -
>
>     Echo path length is 128 ms and all the other settings are default
>     -
>
>     Echo path length is 256 ms and all the other settings are default
>     -
>
>     Camera codec is H264 (codecType=264) and all the other settings are
>     default
>     -
>
>     Camera codec is H263 and all the other settings are default
>
> There are several other settings which can affect the sound quality,
> however I would suggest try only AEC and codec for now as these are new
> features implemented in the new release.
>
> In the room, users should check the following things:
>
>     -
>
>     Speak and check if there is echo or not
>     -
>
>     Measure approximate delay (if any) between the video and sound. For
>     example, somebody count to 10 and lifts his fingers, others look for the
>     video and sound synchronization.
>
> To check the echo better, we need several attendees with notebooks - my
> experience shows that echo problems appears with notebooks very often.
>
> What do you think?
>
> Any ideas are welcomed!
>
> Best regards,
>
> Irina.
>
>
>
>
> On Tue, Apr 23, 2013 at 3:53 PM, seba.wagner@gmail.com <
> seba.wagner@gmail.com> wrote:
>
>> IMHO having yet another checkbox where the user can decide to enable or
>> disable AEC can be only an additional thing on top of a general system wide
>> configuration.
>>
>> The problem with this is for example that the average user does not even
>> know what AEC means.
>> Or what is the difference between hardware echo cancellation and software
>> echo cancellation ?!
>> So what would be the benefit for him in having a checkbox tagged "enabled
>> AEC" ?
>>
>> One of the concerns people have against OpenMeetings is its complexity.
>> They just want to conference and not bother around with a lot of options.
>>
>> I would also not disable the AEC feature by default as long as there is no
>> concrete use-case where AEC (or enhancedMicro mode) leads to an issue.
>>
>> Same for h264. There was a discussion that included h264 a couple of weeks
>> ago:
>> http://markmail.org/message/fctg6eo5qf5hhcaa
>>
>> The consens was:
>>   - The amount of CPU that our clients needed more was not measurable.
>>   - There is no concrete use case where enabling or disabling h264 would
>> improve anything
>> I agree that having a config for the Video codec, same as the Microphone
>> Mode, would be nice.
>> But changing any defaults now back and forth without concrete use cases and
>> some kind of test plan does not make a lot of sense to me.
>>
>> So what I would propose is:
>>   - Make Microphone and VideoCodec configurable and documentate this
>>   - Set up a concrete test plan and communicate that through our mailing
>> lists.
>> Based on the feedback we might then discuss changing some defaults.
>>
>> Sebastian
>>
>>
>>
>>
>>
>>
>> 2013/4/23 Maxim Solodovnik <solomax666@gmail.com>
>>
>>> My previous comment was regarding video codec
>>> Auto selecting h264 based on SIP setting is not a good idea IMHO
>>>
>>>
>>> On Tue, Apr 23, 2013 at 3:17 PM, Artyom Horuzhenko <akhor666@gmail.com
>>>> wrote:
>>>> Usually (for example, in Linphone) echo cancellation option is
>> available
>>>> for every user. In my opinion it would be better to allow users to
>> decide
>>>> use echo cancellation or not because its efficiency depends on
>> hardware.
>>>>
>>>> 2013/4/23 Maxim Solodovnik <solomax666@gmail.com>
>>>>
>>>>> Maybe it is better to make it configurable somehow? config.xml is
>> good
>>>>> candidate IMHO
>>>>>
>>>>>
>>>>> On Tue, Apr 23, 2013 at 2:56 PM, Artyom Horuzhenko <
>> akhor666@gmail.com
>>>>>> wrote:
>>>>>> Hello,
>>>>>>
>>>>>> I suppose using h264 would be necessary only in SIP rooms, for all
>>>> other
>>>>>> rooms - codec which used before (h263?). It avoids any suspections
>>>> about
>>>>>> h264 issues and increses performance. I'm also going to implement
>>>>> switching
>>>>>> microphone mode in the audio/video test dialog. Is there anybody
>> who
>>>>>> disagree with this idea?
>>>>>>
>>>>>>
>>>>>> 2013/4/23 Alexei Fedotov <alexei.fedotov@gmail.com>
>>>>>>
>>>>>>> That would be great!
>>>>>>> On Apr 23, 2013 5:36 AM, "Maxim Solodovnik" <
>> solomax666@gmail.com>
>>>>>> wrote:
>>>>>>>> @Alexei this definitely can be done in realtime, BUT ...
:)
>>>>>>>>
>>>>>>>> The setting affects Working microphone so it is impossible
to
>>>> ON/OFF
>>>>> it
>>>>>>>> while speaking :(
>>>>>>>> The only option I can see is to add dropdown control to
>>>>> testSetupDialog
>>>>>>>> (the one with choose camera and test recording). So we can
have
>>>>>> something
>>>>>>>> like:
>>>>>>>>
>>>>>>>> 1) default system setting (can be moved to Admin->configuration
>>> to
>>>>>> avoid
>>>>>>>> rewriting it after each upgrade)
>>>>>>>> 2) per user setting stored in LocalData (like rememberLogin
>>> default
>>>>>>> cam/mic
>>>>>>>> etc.)
>>>>>>>>
>>>>>>>> using above scheme user can change the setting while entering
>> the
>>>>> room
>>>>>> or
>>>>>>>> "re-sync" audio/video
>>>>>>>>
>>>>>>>> What do you think?
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> On Tue, Apr 23, 2013 at 3:24 AM, Joseph Karwat <
>>>> jkarwat@jellnet.com>
>>>>>>>> wrote:
>>>>>>>>
>>>>>>>>> We tested AEC out-of-the-box. We just installed and ran
a
>>> meeting
>>>>> in
>>>>>>>>> OM2.1- Litle or no echo.  We then, using the exact same
>>>>>> configuration,
>>>>>>>>> logged in to an OM2.0 server and ran a meeting. We
>> experienced
>>>>>>>> substantial
>>>>>>>>> echo. We were both running a PC with Flash echo cancellation
>> on
>>>> but
>>>>>> no
>>>>>>>>> hardware echo cancelation.
>>>>>>>>>
>>>>>>>>> Also for OM2.1 Where do I set the buffer lenghth?  Are
there
>>> any
>>>>>> other
>>>>>>>>> parameters/options I  should know about?
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> Joseph Karwat | CEO | 415-462-0263 | JellVideo |
>>> www.jellnet.com|
>>>>>>>>> JELL NETWORKS, INC.
>>>>>>>>> Enterprise Video Collaboration Solutions
>>>>>>>>>
>>>>>>>>> Click Here to Register for your JellVideo Collaboration
>> Account
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> -----Original Message-----
>>>>>>>>> From: Alexei Fedotov [mailto:alexei.fedotov@gmail.com]
>>>>>>>>> Sent: Monday, April 22, 2013 12:49 PM
>>>>>>>>> To: dev@openmeetings.apache.org
>>>>>>>>> Subject: RE: AEC in the new release
>>>>>>>>>
>>>>>>>>> Joseph, which devices show improvement? Which buffer
length
>> did
>>>> you
>>>>>>> use?
>>>>>>>>> I like the suggestion - Maxim, how is it possible to
>> configure
>>>> AEC
>>>>> in
>>>>>>>>> realtime?
>>>>>>>>> On Apr 22, 2013 11:30 PM, "Joseph Karwat" <
>> jkarwat@jellnet.com
>>>>>> wrote:
>>>>>>>>>> We have a test OM2.1 server installed at AWS running
>> Unbuntu
>>>>> 12.04.
>>>>>>>>>> We tested AEC this morning for the first time and
it did
>> seem
>>>> to
>>>>>> make
>>>>>>>>>> the audio work very well.  Little if no echo.  We
still
>> need
>>> to
>>>>>> test
>>>>>>>>>> under a wide range of scenarios, but we are encouraged
by
>> the
>>>>>>>>>> performance this morning.
>>>>>>>>>>
>>>>>>>>>> If you do make it a configurable setting. It should
be an
>>>> option
>>>>>> when
>>>>>>>>>> setting up the room or allow the individual user
to
>>>> enable/disabe
>>>>>> AEC
>>>>>>>>>> when in a conference room.
>>>>>>>>>>
>>>>>>>>>> Making it a system-wide configuration option will
delay
>>> testing
>>>>> and
>>>>>>>>>> adoption.
>>>>>>>>>>
>>>>>>>>>> We will be conducting further testing of AEC over
the next
>>> few
>>>>> days
>>>>>>>>>> and will report back with our findings.  Anyone wishing
to
>>>>>>> participate
>>>>>>>>>> in our AEC testing is encouraged to contact me.
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> Joseph Karwat |415-462-0263 | JellVideo | www.jellnet.com|
>>>> JELL
>>>>>>>>>> NETWORKS, INC.
>>>>>>>>>> Enterprise Video Collaboration Solutions
>>>>>>>>>>
>>>>>>>>>> Click Here to Register for your JellVideo Collaboration
>>> Account
>>>>>>>>>>
>>>>>>>>>> -----Original Message-----
>>>>>>>>>> From: Alexei Fedotov [mailto:alexei.fedotov@gmail.com]
>>>>>>>>>> Sent: Monday, April 22, 2013 12:49 AM
>>>>>>>>>> To: dev; user@openmeetings.apache.org
>>>>>>>>>> Subject: Re: AEC in the new release
>>>>>>>>>>
>>>>>>>>>> There are no answers. My point here is that we can
be 100%
>>> sure
>>>>>> that
>>>>>>>>>> today AEC does not work as expected, and provides
>> distinctive
>>>> and
>>>>>>> high
>>>>>>>>>> quality echo in 100% of situations when users use
separate
>>>>> dynamic
>>>>>>> and
>>>>>>>>>> microphone systems.
>>>>>>>>>>
>>>>>>>>>> The new thing is AEC mechanism sometimes meddles
with
>>> hardware
>>>>> echo
>>>>>>>>>> cancellation support providing temporary sound outages.
It
>>> also
>>>>>>>>>> provides visible de-synchronization between sound
and video
>>>>>> channels.
>>>>>>>>>> If someone has a different experience, please, speak
up.
>>>>>>>>>>
>>>>>>>>>> So I have two ideas to speak of. 1. Disable AEC in
>>> openmeetings
>>>>>>>>>> releases by default. It takes a simple configuration
change
>>> to
>>>>>> enable
>>>>>>>>>> it. 2. Later understand when AEC really works, and
fix it.
>>>>>>>>>>
>>>>>>>>>> Folks, please again let me know if you disagree and
have
>>>>> beneficial
>>>>>>>>>> AEC experience.
>>>>>>>>>> --
>>>>>>>>>> With best regards / с наилучшими пожеланиями,
Alexei
>> Fedotov
>>> /
>>>>>>> Алексей
>>>>>>>>>> Федотов, http://dataved.ru/
>>>>>>>>>> +7 916 562 8095
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> On Thu, Apr 18, 2013 at 11:11 PM, Alexei Fedotov
>>>>>>>>>> <alexei.fedotov@gmail.com>
>>>>>>>>>> wrote:
>>>>>>>>>>> Hello guys,
>>>>>>>>>>> How AEC (acoustic echo cancellation) works for
you in the
>>> new
>>>>>>>> release?
>>>>>>>>>>> Can anyone share usage experience?
>>>>>>>>>>>
>>>>>>>>>>> --
>>>>>>>>>>> With best regards / с наилучшими пожеланиями,
Alexei
>>> Fedotov
>>>> /
>>>>>>>>>>> Алексей Федотов, http://dataved.ru/
>>>>>>>>>>> +7 916 562 8095
>>>>>>>>>>
>>>>>>>>>
>>>>>>>>
>>>>>>>> --
>>>>>>>> WBR
>>>>>>>> Maxim aka solomax
>>>>>>>>
>>>>>
>>>>>
>>>>> --
>>>>> WBR
>>>>> Maxim aka solomax
>>>>>
>>>
>>>
>>> --
>>> WBR
>>> Maxim aka solomax
>>>
>>
>>
>> --
>> Sebastian Wagner
>> https://twitter.com/#!/dead_lock
>> http://www.webbase-design.de
>> http://www.wagner-sebastian.com
>> seba.wagner@gmail.com
>>


Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message