Audio

From Mumble Wiki
Revision as of 16:18, 1 June 2020 by Toby63 (talk | contribs) (Echo Cancellation: smaller changes to the text, for better readability)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search
Icons oxygen 48x48 actions view-pim-notes.png
Note:
This page is missing some parts and might be extended in the future.

Introduction

Mumble features various Audio Options and Features, this wiki page will describe them and tell you which settings are best for your setup.

Audio Input - Options

Interface

Echo Cancellation

If enabled, this will filter echo from the audio you send to others.

You should enable it in the following cases:
1. You use conventional loudspeakers and no headset:
This Setup will usually create echo, so you will need echo cancellation.

2. Headset transmits echo, though it shouldn't:
Some Headsets tend to transmit echo, the reasons for this include: bad cable and noise isolation or the microphone is to close to the loudspeakers of the headset.

Two options are implemented (you can choose one of them):

Option: Description: Usecase:
1. Mixed echo cancellation This is the basic Option: It will process all loudspeaker outputs bundled together.
This is less accurate than the Multichannel option, but will also use less CPU.
Sufficient for setups with loudspeakers near to the microphone.
2. Multichannel echo cancellation Extended option: It will process all audio channels seperately, this is more accurate, but will result in higher CPU usage. For setups with (multiple) loudspeakers farther away from the microphone.

In case of doubt:
1. Try Mixed echo cancellation first.
2. If Mixed echo cancellation does not filter the echo correctly, try Multichannel echo cancellation.


Alternatives:
You can also use external echo filters.

Icons oxygen 48x48 actions view-pim-notes.png
Note:
In this case you should deactivate the echo filters in Mumble (to avoid interference).

Examples for external echo filters:

Audio Processing

RNNoise

RNNoise is a noise suppression library.
It is intended to filter background noises from the audio that is sent to other users.
RNNoise uses a deep learning algorithm so it should work better than most regular filters.

More information can be found on various websites:


Audio Output - Options

Technology

In this section we take a short look at the technology behind the scenes.