Noise canceling feature available?

hi, i am new to jitsi and installed it on one of my machines for testing purposes (maybe to host a little server/instance of jitsi for my students). unfortunately, the background noise is gigantic.
when i use zoom, teams (or any other conferencing software i know) i never heard the background noise in such detail. it really destroys every conference and unfortunately, thats why jitsi is not usable for me.

i already tried some of the flags ( disableAEC, disableAGC, disableAP, disableHPF, disableNS) but none of them helped in a way that i would say it is useful.

would it be possible to somehow configure the web client in such way, that a special level of noise must be reached before it is send, like teamspeak provides? i mean, it does not need to be some special high end AI but a simple volume meter would be really great i guess!

edit: i found about 20 other topics with a similar or the same problem but i can not post more than 2 urls here :confused:

‘DisableNS’ (that means: disable Noise Suppression) is only a hint that the server sends to the browser, and the browser does what it wants with it. If the browser don’t find on the operating system side any help to do noise cancellation, there is nothing more that it can do and there is nothing more that Jitsi-meet will do, the whole ‘handling’ of media server side by Jitsi meet is routing. Nothing more. No change of media is done on the server ever.
DisableNS is disabled by default (that is, noise cancellation is enabled by default) and setting DisableNS to true has only one usage: if noise cancellation is somehow bad for the use case (maybe music ?), Jitsi-meet allows you to disable it for the meetings. That’s it and that’s generally not useful.

So if your audio driver supports filtering of sound other than voice it may work. Most of the time such features if available are much too complicated to be setup by users so the only reliable option is to disable mics by default. That’s the price of using a generic software such as a browser for client.

thanks for clarification.
just curious how ms teams works without problems (in the same browser). maybe they have some db filter in it which activates the routing.
i am not asking for a high end solution, but some very simple filter possibility would be awesome like “only send if the sound is louder than x db”.
in german there is a saying like “ich hör die flöhe husten”, and thats what is happening now :confused:

how could I know ? Maybe it’s some stuff programmed specifically in Edge.

there is a sound level setting in most desktop OSes.
here is mine:

Here the level (‘volume d’entrée’) is at a rather low level; if I touch lightly the mic, there is nothing seen in the meter (‘niveau d’entrée’). If I raise the level to 2/3 and brush the mic however softly, something appears in the meter.

Theoretically there is nothing stopping the browser to be able to manager these parameters out of the box or even surfacing it to the app (in this case it would be Jitsi-meet javascript). But in my case there is no parameter I see in my chromium build so I assume that it is not generally doable (that the issue with Jitsi-meet they are limited to the features common of at least of Chromium based browsers, Safari and IOS browsers - and possibly Firefox)

i use chromium on fedora 35

looks like pavucontrol, i am running fedora 35 with pipewire, i also have something like that but unfortunately, i can not control when the system should send the signal and when not. the only thing i can do in this dialog is to control how intense the signal should be.

That’s basically the same control. In my pavucontrol version, at right end of what is called for you ‘Eingangsniveau’ there is the actual percentage and decibel and it seems to be lacking in your case, so it’s difficult to do a fair comparaison; but from what I see in my version, your sound level is set to a very higher level than mine.

As we are also facing same issue.Did we find some option that works?