Jitsi meet for music teaching

Is there any interest in Jisti meet for codecs more suitable for music instrument teaching? In my world of lockdown, all instrumental teaching has gone online and the results in audio are poor to hopeless.
The lessons with any student above beginners need to take in to account quality of tone and dynamic range - both of low consideration in a video conference setting.
I have searched the forum but have obviously not found what I need.
Instrumental lessons from professional teacher players need controls for switching automatic gain control off, as well as much switchung in higher data rates for the audio stream (stereo is unimportant).
The well advertised popular video programmes have some audio controls who’s results are deeply unimpressive.

Any thoughts?

Michal

Having music-quality audio in Jitsi is very hard to achieve, as you probably already saw in the several other threads here about it. I would love to be able to use Jitsi for band rehearsals, for example, and it’s not really possible - even though this is lower requirement than the music teaching you mention.

The reason is that Jitsi is built around a different concept - to get a really high audio (same for video, if that matter) quality it’s needed to rely on server side mixing, and Jitsi as a platform goes the opposite way. It forwards all the media, instead of wasting huge server resources to mix them. While audio could be mixed, it’s a lower quality and it doesn’t load the platform much - if the audio was very high quality, it would beat the purpose of Jitsi Meet to mix it.

I believe there could be improvements in that area, but currently you can’t have professional music-grade audio. What you can try is to use two programs and two connections - one for video, through audio-muted Jitsi Meet and the second one (in a separate app) for audio, using the audio platform that suits your use case best - something based for example on Jamulus, Ninjam, etc.

I think there are some threads here for integrating Jitsi with Jamulus (and similar), you can check them out. That said, my personal experience was not best and I haven’t seen someone share positive experience here. (Good) audio is always a problem. Of course, I’ll be happy to be proven wrong.

Welcome to the forum, @MichalH

@yasen is right that Jitsi was not developed specifically for music teaching in mind, BUT all hope is not lost really. You can disable audio processing in Jitsi by activating some of the properties in config.js. Here are the available options:

    // Disable measuring of audio levels.
    disableAudioLevels: false,
    
    // audioLevelsInterval: 200,

    // Disables Acoustic Echo Cancellation
    disableAEC: false,

    // Disables Automatic Gain Control
    disableAGC: false,

    // Disables all audio processing
    disableAP: false,

    // Disables Highpass Filter
    disableHPF:false,

    // Disables Noise Suppression
    disableNS: false,

    enableLipSync: false,

    stereo: true,

Adjust the properties to suit your requirement.

While mixing sound at the server level is indeed possible without huge resources, it’s not done by Jitsi as far as I know.

Yes, I meant to write “while it could be mixed”, I’m editing my post now. Thanks for noticing.

I tried several configurations playing with these audio options, but I am struggling to get audio quality close or same as Zoom.

My main issue is with distorted guitar sounds, specially when mixed with playbacks (i.e., sending both the guitar and a playback together, already mixed, to the input). The sound gets a flanger / wah-wah effect probably due to the system identifying some sounds as noise to be supressed.

If I disable everything, the sound is great, but then both student and tutor need to use headsets, which is not always possible unfortunately.

My question is: how can zoom cancel noise and echo and still not generate this horrible effect? Is this something Jitsi still can’t replicate?

Of course if I enable the “true original sound” on Zoom, then it also generates echo (I can hear myself from the outputs of the student), but even with the true original sound disabled, and low echo cancellation enabled, Zoom does not generate the strong supressions killing the sound of playbacks and/or distorted guitars.

:slight_smile:

You shouldn’t disable everything; you have to play with the values and find which works best for you. AEC for instance, should be enabled to avoid echoes. I’ve used Zoom and I use Jitsi; I can tell you categorically then even when it comes to audio, Jitsi outshines Zoom.

What would you recommend?

AEC enabled only?

My problem is just with the echo and with the “degradation” when some processing kicks in during playback of backing tracks or distorted guitar.

I think you might also want to disable AGC. Again, my recommendation is that you play with those flags till you get exactly what you want. It’s in there, I promise.

1 Like