logo

Live Production Software Forums


Welcome Guest! To enable all features please Login or Register.

Notification

Icon
Error

Options
Go to last post Go to first unread
MattiaCC  
#1 Posted : Friday, May 26, 2017 11:08:42 AM(UTC)
MattiaCC

Rank: Member

Groups: Registered
Joined: 6/14/2016(UTC)
Posts: 12
Location: Italy

Hello, I'm in my first vMix video call and during the test with other remote guest everythink works fine but now, the audio from the client has a bad quality and isn't continuos, every 5/6 second the audio go low and dropped out then it come back higher.

Is a low bandwich issue? The video is ok and the problem is onlu on the audio side.

The client see and ear me with no problem, in my building we have a fiber connection but from the client side they have only a slow connection (20Mbps down and 1Mbps up)

Thank You
MattiaCC  
#2 Posted : Friday, May 26, 2017 1:06:16 PM(UTC)
MattiaCC

Rank: Member

Groups: Registered
Joined: 6/14/2016(UTC)
Posts: 12
Location: Italy

Ok, I've found the problem, the echo canceller built in the vMix software cutout the guest voice everytime the electronic gate of the local radio microphone capture the room speaker.
I'll made some other test.
Anyone with the same experience?
mjgraves  
#3 Posted : Friday, May 26, 2017 1:46:44 PM(UTC)
mjgraves

Rank: Advanced Member

Groups: Registered
Joined: 7/1/2015(UTC)
Posts: 1,151
Man
United States
Location: Houston TX

Thanks: 319 times
Was thanked: 263 time(s) in 233 post(s)
vMix Call leverages WebRTC. The browsers themselves have echo cancellation, automatic gain control and noise reduction functions.

The performance of these functions varies between browsers. The echo canceler in Chrome is not very good. Under normal circumstance you won't notice it's major fault. However, if you have people talking at the same time you might notice the problem.

Over a webrtc connection in Chrome the outgoing audio (microphone) is mangled when there's also incoming audio (to the speakers.) The circumstance is called "Double-Talk." There's undesirable interaction between the in/out streams due to faulty EC processing.

An entirely new echo canceler (AEC3) is implemented in the coming release of Chrome (59.) So there's hope that not too far down the road the EC in Chrome will improve.

All of that may have nothing to do with the problem you face. It's just worth noting that the audio processing in the browser is largely beyond the control of vMix. They can only turn it on/off using flags in the code at vmixcall.com.
sunkast  
#4 Posted : Friday, May 26, 2017 2:39:47 PM(UTC)
sunkast

Rank: Advanced Member

Groups: Registered
Joined: 3/7/2017(UTC)
Posts: 60

Thanks: 3 times
Was thanked: 6 time(s) in 6 post(s)
mjgraves wrote:
vMix Call leverages WebRTC. The browsers themselves have echo cancellation, automatic gain control and noise reduction functions.

The performance of these functions varies between browsers. The echo canceler in Chrome is not very good. Under normal circumstance you won't notice it's major fault. However, if you have people talking at the same time you might notice the problem.

Over a webrtc connection in Chrome the outgoing audio (microphone) is mangled when there's also incoming audio (to the speakers.) The circumstance is called "Double-Talk." There's undesirable interaction between the in/out streams due to faulty EC processing.

An entirely new echo canceler (AEC3) is implemented in the coming release of Chrome (59.) So there's hope that not too far down the road the EC in Chrome will improve.


I did a little bit of reading about this and one thing to note is that while AEC3 will be improved in 59, it will not be on by default. You will have to enable it by changing a flag.
thanks 1 user thanked sunkast for this useful post.
sinc747 on 4/5/2018(UTC)
mjgraves  
#5 Posted : Friday, May 26, 2017 2:43:43 PM(UTC)
mjgraves

Rank: Advanced Member

Groups: Registered
Joined: 7/1/2015(UTC)
Posts: 1,151
Man
United States
Location: Houston TX

Thanks: 319 times
Was thanked: 263 time(s) in 233 post(s)
sunkast wrote:
I did a little bit of reading about this and one thing to note is that while AEC3 will be improved in 59, it will not be on by default. You will have to enable it by changing a flag.


Yes, that's true. It will require that devs change their code. It won't be in widespread use until release 60 or later.
MattiaCC  
#6 Posted : Friday, May 26, 2017 4:58:12 PM(UTC)
MattiaCC

Rank: Member

Groups: Registered
Joined: 6/14/2016(UTC)
Posts: 12
Location: Italy

Thank you for the information, we already use webRTC with a voip phone system for video collaboration and know that this protocol is promising but has some limitation like the QoS problem and the Microsoft/Apple incompatibility.
In my today's use I'm in a meeting room and the two radio microphone is moving between the audience for the question, literally under the ceiling speaker, probably the worst situation.

We also try with another videocall system hosted by the guest who send us an invitation link that works only under Internet Explorer and need a plugin installation (so not a webRTC system), in this test the echo cancellation works better but the audio and video quality is really worse so the event manager prefer the quality of vMix call and all the stuff that virtual set and merge function can offer ;-)

So, let's wait and hope for some performance implementation from webRTC development and a big thanks to the vMix staff for the really appreciated new videocall feature!!
ask  
#7 Posted : Friday, May 26, 2017 10:15:03 PM(UTC)
ask

Rank: Advanced Member

Groups: Registered
Joined: 10/13/2012(UTC)
Posts: 1,162
Australia
Location: Melbourne

Thanks: 220 times
Was thanked: 199 time(s) in 181 post(s)
MattiaCC wrote:

In my today's use I'm in a meeting room and the two radio microphone is moving between the audience for the question, literally under the ceiling speaker, probably the worst situation.


What you are describing here is a feedback loop. This a issue that will not be solved by software. This requires you to know and choose your equipment carefully. My guess is that your radio mike is omni-directional. I.e. picks up sound from multiple directions, and may even be a condenser microphone. In that environment, look for one that is uni-directional with good rear-rejection, is dynamic and then get it as close to the speaker as possible. Keep the return sound (from the PA) as low as possible.

You then want to tune the room by using a separate eq device to reduce the cross-talk as much as possible, and also perhaps look at also using a feedback eliminator on the PA. You can also try a noise gate.



MattiaCC  
#8 Posted : Saturday, May 27, 2017 4:48:55 AM(UTC)
MattiaCC

Rank: Member

Groups: Registered
Joined: 6/14/2016(UTC)
Posts: 12
Location: Italy

No, no, I'm probably be misunderstood, it isn't a feedback loop, the system has a digital processing core for the room with gate, equalization, and correct routing for the vMix call I/O, also the radiomic works very well without feedback problem.
Like I wrote we made a test with other video call system, non webRTC based and it's works better without eco cancellation problem, finally I think that it is a bad quality alghoritm of the webRTC protocol under Chrome.
The next week I'll made some test with an other webRTC based system, just for curiosity, in comparison with vMix Call, using firefox and chrome.
Thanks for the interesting and sorry for the bad english.
DWAM  
#9 Posted : Saturday, May 27, 2017 6:31:10 AM(UTC)
DWAM

Rank: Advanced Member

Groups: Registered
Joined: 3/20/2014(UTC)
Posts: 2,721
Man
France
Location: Bordeaux, France

Thanks: 243 times
Was thanked: 794 time(s) in 589 post(s)
Ciao Matia

do not make definitive conclusions out of a single test.

I suggest you make a vMixCall from a remote computer on your local network with a good mic and a guest setup you are sure of.

If the same issue happens again, then we'll see. But I feel like your issue is related to your host setup, not the guest's (you said he could hear you well)

Also you said "the echo canceller built in the vMix software cutout the guest voice everytime the electronic gate of the local radio microphone capture the room speaker."

But there is no EC on the vMix side, vMix only does a mix minus. This issue is more likely related to your radio setup than vMix. As Ask told you it is a feedback loop eliminator handled by your radio equipments, not vMix. Suggestion made to use cardiod mics instead of omni-directional is correct, or simply use headphones instead of loudspeakers in the room to confirm what we're saying.

Guillaume
mjgraves  
#10 Posted : Saturday, May 27, 2017 10:21:42 AM(UTC)
mjgraves

Rank: Advanced Member

Groups: Registered
Joined: 7/1/2015(UTC)
Posts: 1,151
Man
United States
Location: Houston TX

Thanks: 319 times
Was thanked: 263 time(s) in 233 post(s)
Millions of people use webrtc-based services daily. They work, and some work very well. I'd suggest that you try one or two of the handful of reference standard tools. For example, Talky.io, Jitsi Video Bridge, or app.rtc.

app.rtc in particular allows you to modify the performance of the system, so you can add/remove EC on an experimental basis.
MattiaCC  
#11 Posted : Tuesday, June 6, 2017 6:28:54 PM(UTC)
MattiaCC

Rank: Member

Groups: Registered
Joined: 6/14/2016(UTC)
Posts: 12
Location: Italy

Here I am, again, today I have made more test, different equipment and different location. One laptop with vMix software wich host a call from another pc who use Firefox.
I try with vMix and also with other two webRTC client, 3CX WebMeeting and appr.tc.
After many test I find this: the two web rtc solution works fine with no problem, instead, using vmixcall, it works fine only if I use a couple of headphones but if the speaker volume is high and the microphone capture the voice of the callers the volume go lower for a while than turn up. This problem doesn't occur with the other two webrtc solution.
I think that the audio mix minus that vmix made go to cancel the audio who is coming from the callers, is quite simple to chek by send a continuos note from the callers to the vmix, the audio level will go down and up, with other webrtc test the audio level remain high.

I hope that the explain is understandable.
DWAM  
#12 Posted : Tuesday, June 6, 2017 7:21:19 PM(UTC)
DWAM

Rank: Advanced Member

Groups: Registered
Joined: 3/20/2014(UTC)
Posts: 2,721
Man
France
Location: Bordeaux, France

Thanks: 243 times
Was thanked: 794 time(s) in 589 post(s)
It is not recommended to use loudspeakers next to microphones, and not only for vMixCall, but all things production.
Better use headphones if you want to PA your production or use cardoid microphones which won't pick up ambient sounds.

As of comparing vMix and WebRTC services, they do not work the same way for sound: WebRTC services do Echo Cancellation while vMix does Mix-Minus. This is logical considering each specific use.
thanks 1 user thanked DWAM for this useful post.
kl_clips on 1/23/2018(UTC)
mjgraves  
#13 Posted : Tuesday, April 3, 2018 10:48:27 AM(UTC)
mjgraves

Rank: Advanced Member

Groups: Registered
Joined: 7/1/2015(UTC)
Posts: 1,151
Man
United States
Location: Houston TX

Thanks: 319 times
Was thanked: 263 time(s) in 233 post(s)
DWAM wrote:
As of comparing vMix and WebRTC services, they do not work the same way for sound: WebRTC services do Echo Cancellation while vMix does Mix-Minus. This is logical considering each specific use.


I don't know this for a fact, but it's highly likely that vMix Call leaves the browser EC/NC/AGC enabled. These are the defaults. Without EC, if a guest was not using a headset, just the built-in mic/speakers on their laptop, you would get a nasty result even with the far-end mix-minus.

This should be pretty easy to confirm using Chrome:///webrtc-internals while connected to a vMix call from a client. That would expose the GetUserMedia params.
thanks 1 user thanked mjgraves for this useful post.
DWAM on 4/3/2018(UTC)
Users browsing this topic
Guest (3)
Forum Jump  
You cannot post new topics in this forum.
You cannot reply to topics in this forum.
You cannot delete your posts in this forum.
You cannot edit your posts in this forum.
You cannot create polls in this forum.
You cannot vote in polls in this forum.