Simultaneous interpretation


We have some clients which are currently using the “simultaneous interpretation” feature of ZOOM to manage with audio translations in their webinars (

They are moderately satisfied with this solution and really need an Open Source one : Do you think it can be possible to integrate this kind of “all-in-one” conference feature in JITSI ?

Does anyone have already test something for that, or does we have ideas ?

Thx !


Jigasi supports captions using google-speech-to-text service, it also supports translating this captions, there is just no UI implemented for that.
About translating the audio not sure, but I suppose it is doable … the problem is that this brings delay and will de-sync quite a bit the translation and the real audio.

Thx for your answer Damencho : the purpose here is not to translate the audio to text but to have real translators in the jitsi room, and allow each participants to choose which one they want to hear.

Basicaly, if there is a webinar between 1 polish and 1 spanish guy and if want to propose an english and a french traduction, I will also have in the jisti room :

  • 1 polish to french traductor
  • 1 spanish to french traductor
  • 1 polish to english traductor
  • 1 spanish to english traductor

The purpose here is to allow each participant to “activate” the french or english translation by clicling on a “translate” button which will activate the audio for both translaters of the chosen language (and reduce the volume of the polish and spanish guy).

I hope my request is more clear :slight_smile: !

Anyone ?

Our organization is also very interested in a simultaneous interpretation option in Jitsi. We have used the interpretation tool a lot in Zoom. But since Zoom is a US company, it is not possible to use it for communications with some countries (for example we work with labour organizations in Cuba, but Zoom cannot be used there due to the embargo), and in any case, we prefer to use open software. If anyone has managed to develop a similar toolfor Jitzi, we’d love to hear about it.

This can be done manually using the volume slider for each user. If there is only 4 active speakers in a room, not a bad choice…

This can be done manually using the volume slider for each user. If there is only 4 active speakers in a room, not a bad choice…

Yes… but no :slight_smile: !

The purpose is not to find a workaround for one simple case with the existing features, but to think about a global solution that respond to any needs.

For exemple, if I made a webinar with 5 experts which each speak in a different language (for exemple : polish, russian, french, italian, greek), so for a traduction language I must have :

  • 1 polish to target language traductor
  • 1 russian to target language traductor
  • 1 french to target language traductor
  • 1 italian to target language traductor
  • 1 greek to target language traductor

(Note : If the target language is polish, there is no need to have a “polish to polish” traductor, so there will be only 4 traductors)

So if I want to offer a polish, russian, french, italian, greek and also english traduction I must have 25 traductors (it’s how it works during international meetings).
(Note : 25 = 4 traductors for each 5 expert language + 5 traductors for english)

In this conditions, I can’t deal with manual volume sliders in a jitsi room with about +30 peoples…
(Note : and “+30” is only for the experts, traductors and administrators of the webinar, but what about the xxx viewers ?)

Since it’s possible to do manually, it can be automated for larger group.

Probably it will be needed a customized login panel which creates a JWT token which contains the language info for each participant and some customizations on the UI which manage volumes according to the JWT context

Right, a customer has requested a rate for such a feature.

For another customer, we’ve implemented a sign language interpreter feature, which we’d be pleased to be merged into Jitsi-meet:

@emrah @damencho Fancy it?



This is nice. Good job, @kpeiruza.
I’d be interested to know how you got the interpreter’s video feed. Or did you use OBS?

We implemented it on jitsi-meet interface.

A user with a special prefix at its name is shown in that position.

Best regards!

El dv., 29 de gen. 2021, 20:19, Freddie via Jitsi Community Forum - developers & users <> va escriure:

1 Like

very nice :+1:

How can I use this? What is the prefix?

I’m in the same position. Any progress on getting a feature like this in Jitsi?

It would be awfully useful to me as well. Is there any news?

1 Like

I built a frontend that does this a year ago. I’ll have to update it because it’s currently too out of date to run on current browsers. but here is how the one I built works.Simultaneous translation frontend - YouTube

Unless you open-source your solution, it’s probably not of much use to people here seeking to build on their own deployments.

I found a solution that seems promising. It uses JS and the api to enable volume adjustment between the main feed and a user with “interpreter” in their name, plus a few other functions like a “slow down” request:

The interpreter must include the word “interpret” in their display name. When the interpreter hears language A, they interpret into language B. When the interpreter hears language B they interpret into Language A. Multiple interpreters can be on the call and trade on and off interpretation duty.

When the program runs, it pulls in your Jitsi Meet instance via an iframe. When each participant changes their slider, it adjusts the local volume of all participants based on whether or not they have the word “interpret” in their display name.

The limitations are that there can be only one language pair, that it only works in browsers and that it doesn’t work at all on iOS devices, because Apple doesn’t allow JS volume manipulation. I have installed it on our server and it seems to be working but haven’t tried it in practice yet, plus its usability in my case is limited as the problem with iOS can be a substantial inconvenience for some people in my organisation.

Here is the repo: May First Movement Techology / jsi · GitLab, a working instance provided by the author’s (Profile - jamie - Jitsi Community Forum - developers & users) organisation, and a thread on the forum: Adjust volume levels via javascript.

1 Like