Jamulus VST3

[AndyMcProducer]

Hiya, so I have a vst3 version in dev of Jamulus still using qt wrapped in juce.
Using AI to help get it there.
Once I got the VST3 100% and using everything possible from the standalone I#ll ask for help on putting it up. like commit it.
Before that I will do other versions like lv clap and au but I'd need others on those platforms with dev knowledge to test and help fix any teething probs.
I tried a few ways, to convert it to juce but then some things wouldn't be as is and needed converting, possibly to a detriment of performance,. I tested a few methods not needing QT as the QT version for this is wanting 5.7.2 and can't even get that down now so I up'd it to use a newer v6 but not latest as I could imaging some things completely changing from 5.7.2 to latest 6.10. and it breaking.
It might arrive with some bugs but I think with the majority done and the fact it is a lib of the source client that peeps here could
possibly fix any issues.

Reason I'd like to have converted it to complete juce is only so much of the dependencies are not needed and fixing, debugging and building
is so much less probs.
Anyway,
Merry Xmas.

[ann0see]

Thanks you very much! Yes, it seems like Qt should be phased out in the future.
We already have a PR on moving to QML (which might fix some UI bugs on Qt6 and mobile) which is something to think about for Jamulus 4.

You could open a PR - however it'll be a very long time until this is merged (if it is even merged). What you could do is to use the auto build progress and link the binaries for people to test.

@pljones already had some thought on Juice, so tagging him here.

[ann0see]

Once I got the VST3 100% and using everything possible from the standalone I#ll ask for help on putting it up. like commit it.

Do you need help with git?

[AndyMcProducer]

yes will do, sometimes when pulling a git it has the commit option their but i didn't pull it liek that, did it differently. Before putting up the git ill clean the code up first but yo guys would need to go on that urther as porbs lots of none needed stuff where things have been attempted and at times AI maybe left crap in their.

[ann0see]

Once you have a repository set up, use git add, git commit and git push to upload your changes.

As reference, something like
https://git-scm.com/cheat-sheet might help.

If you need exact help, please verbalize them here. Otherwise I believe that LLMs will help you.

[pljones]

I have to say I'd rather see this running on CLAP than VST3. Although Steinberg has made it fully open source now, I think we'll see CLAP gaining more traction on more platforms.

[AndyMcProducer]

Should be ok making it clap too Pljones.
I will need help with the git bit possibly.
Also I got it all working witht he QT showing liek in the image above but soooooooooo many probs, maybe newer qt would be better but still.
So I'm keeping atm all the QT backbone but switching the gui from QT to Juce.
When I orig made a few vst3 versions of jamulus it was sorta not great and the server list was such a issue to get converted over.
So this time I've gone another route, got jamulus clones, built it with newer qt, then got all working then stated converting stuff over and
it's been a lot more success.

When you say Jamulus 4, where did 3 go? :)
Also I've added a delay and more settings for the current reverb as they add no latency and did the delay in the same way the verb was done.
Giving it an FX button that pops out the pair with their settings.
Once everything is done then then ill get it up on here. I'll add also ability to have themes folded based and possibly in a simple script so peeps can make their own themes.
Then maybe possibly a theme generator.
But with Juce when ya got one thing sorted the rest like clap lv2 and the rest are much more straight forward.
Like make one and you've made the rest.
But I've kept the orig code and just converted the gui to juce.
I will try later to convert the rest to juce but that wil take some time and be bit by bit buit by then ill have it up on here and you guys can jump on it too.
BTW I'm using AI here too, makes a big diff. Understanding and doing are 2 dif things and even thought it still takes hours and hours with AI
and you have to understand code and what's going on it's mostly failure and wasted hours.
Copilot on vs is free for 30 days and they do actually remind you your trail is ending several times.
Also google.com search ai option which i think ya press tab on the address bar in chrome now can give ya the right direction and tell ya if somit is possible before wasting time attempting it. Gives ya the right direction to go in without all the hassle.
We probs all been their where we start somit then find out where we wanted to go with it isn't possible with our deep OT stuff and AI can help with it and help guide right.

When all the hard long winded hours of wasted tie are done and all is as should be, the rest is like being in a sweet shop and picking ya fav sweets.

[AndyMcProducer]

I just added the dleay and the size ya see is screwed at the bottom but ill fix that soon. Also I added the chat but so it can be popped in and out into it's own window or as a side panel. I didn't know if that was a deliberate thing or not so thought add it then all happy.
Local mixer isn't setup yet or nicer graphics, that's all best done when everything workls then ill sort it so the theme thing can be used.
I did a free bingo caller which I yet t o put up onlien but I added some theme functions to that and also audio files for themes so peeps could liek add a cowboy theme then record all the numbers being called out in a cowboy accent so if they existed it woudl use them but if not it wil use whatever it's running on narriato. Hearing english or other languages in other accents sounds cool. So if on say a fire tc stick it would have alexa calling the numbersand u cna choose liek who ya want from a list box. So was thinking similar idea for jamulus but make it into a script type thing so peeps can set what they want for whatever.
Was thinking also adding cam link button that brings up a cam room per server with the option to also designate one but those things maybe need adding to the server too.
Would it be worth also adding the server into the client too so peeps can create rooms and maybe try see if upnp would work?

[AndyMcProducer]

[AndyMcProducer]

oh also i got it sorting the lanes like for ping server name etc.
When i get it upped on git you guys can add like more stuff if needed. I'm gonna get it to show users in the server when peeps hover over the servr in the list and add a tick box also if clicked it then shows users in each server like it does now.

[AndyMcProducer]

oh also wasn't gonna make it an instrument so has multiple in's and outs but will try and add the sidechain as a input so like you can choose a sidechain input as the audio in source. This will be useful in some situations rather than pull the track audio or master audio.
Also if it goes further later then multiple in's and out's could be used. VST3 supports a lot more than is used atm by what's about.

[AndyMcProducer]

Sorry btw Merry Xmas PLJ, Ann0see. xx

[AndyMcProducer]

Update, connected fine, hear all musicians, chat works, just fine tuning stuff and getting current things working and included form the orig then will added some extra's. Gonna see if opus dred will work too, someone added it to sonobus they they said they can hear a different. Could allow for lower latency settings when cards just struggle a bit with it.
I'm gonna try add some stuff but try get it to not show junk on the current jamulus and only show on like the vst. Mainly to keep backward compatibility and allow new options with out causing issues for previous version.

[AndyMcProducer]

@ann0see @pljones What is groups primarily used for, reason i ask as if it is needed and if is could be better featured, i'd add it when I add the grp option.
Everything else is all working, a bug on closing causes a crash of th ehost but will trace that with some debugging then sort it.
Chat pops in and out and keeps previous chat, has option to save chat log, country banners lay over the users channel strip and the channel changes to the colour of the skill level set. Flags was a paint but they changed the id scheme in v6 of QT but they have a conversion table and that fixed that.
Almos lots loads of stuff done but AI helped me get it all back. phew.
The flag images are a bit low quality for their use but it works but might try get AI to redo them all or try upscale them so they a few more px insize.
Few things I wanna add is a Mono button for every channel, an auto mix/leveler to bring peeps up and down. Also thought about later using mono and levels to do a few auto mix options. Mainly narrowing vocals and mono or narrowing or more to mid on bass and drums and using a db level settings to better mix them in realtime then use some pan or side but this all based on the instrument set. But also thought about a boost button on each channel too, maybe a +6db boost but option in settings to auto level boosted channels so if boosted and it becomes too loud it will take off db's to match other levels or again use the instrument setting.
I added a Monitor button, so peeps can hear local or use delayed latency audio of theirs cos I know some people prefer the delayed sound when playing to keep in time.
You'd think the reverb and delay aren't needed cos it's a daw but some people will want it simple so they can setup their mix in the daw then add verb or delay which I'll try add a few presets to mainly a few for vocals and some a couple defaults. Which makes me thing I should add a width to the verb for vocals.

[ann0see]

What is groups primarily used for

Depends. For my use case it would be a set of instruments which play the same cords.

[AndyMcProducer]

Gonna star on it in a sec, got the auto level sorted. I left attach and release options in their, cos they can help fine tune sometimes but I havbe it keeping and eye on peak levels over a period rather than react straight away. orig i tried to use a attach and release but that sometimes was too choppy so 've kept them for fine tuning but done it so it looks at peak levels over a short period rather than instantly. Works well, will increase those too quiet and decrease those too loud and keeps all in a nice range to prevent master volume clipping. Added a boost button too which will boost someone +6 but these are ignored when auto level is on. Shell I add a short cut too like CTRL held and mouse click on channel which will enable group plus a G button.
So simple CTRL + click sets a quick group option and links faders so u can adjust though grouped faders all at once and then clicking the G button will give groupoptions or maybe a little listbox next to it with say 1-8 in it which probs would work better and be a instant visual indicator. might be better also changing the colour of the G button slightly to also show what group things are in. The current group does that join faders as one cos I'll use the orig code and logic? @ann0see

[AndyMcProducer]

What I did is just hat have a little box open when ya click the G button, but also scrolling mouse wheel on the g button will set and change the group and each have a slightly different colour of green. Faders group fine and don't know if this was in the original but when you had faders grouped and one hit the bottom it got stuck and didn't adjust back up and keep the level ratio it had but does now.

[AndyMcProducer]

@ann0see Ok I think I'm committing it now, if I done it wrong or need to delete or change something and only I can do it form here let me know and I will.
The VST3 is fully working now, still probs could do with some tweaking but I think I've got it mainly bug free.
I've asked Baz to test it out and made a setup for it with inno 6.
I will attempt building the other plugin formats for it tomoz but mac etc I won't be able to test as I have no mac.

I've started implementing a secure method using the current chat to sending encoded hidden messages that other vst atm can see then ask another user to accept liek a PM. from this then later could add video, file sending and possibly even quicker p2p audio if the route was faster than jumping to and form server.
I'm making it this way cos other way would expose peeps ips and this is a very modern way to safely connect decentralised.
It it works out ill put a warning in the pop up for connection, only accept if you know this person. Plus an option to maybe have a friends list and block anyone not on the friends list from sending invites to connect. But no server overhead is needed other than a invite then it handled p2p away form the server so no loading the servers up with more jobs.
If ya wanna test the VST message me, it's not got the p2p invite thing setup in the current build or shoudl have it in the committed one.

[ann0see]

Thank you! It's extremely important that you make small, modular changes as we cannot review a huge PR. Meaning your changes would not be accepted.

So please split the features out in separate pull requests/branches. Also: commit often and explain the changes well!

[AndyMcProducer]

I did but somit happened when making it and I changed IDE and I'm sure I kept changes but then when I built it it built an older version so went back to the other ide and now that was like on the older version. But possibly I still got the previous version and all the commits.
If not though should I push it as a fork and call it JamulusVST? It's up to you guys, can a folk then be pulled back into the main if you guys needed it to?

[AndyMcProducer]

@ann0see Also the main QT should be still intact, this approach was to make the jamulus client into a lib then using juice as a wrapper and for the gui but everything underneath atm is still QT.
I see QT had been updating in the commits then something about it being reverted for mac os?
I did orig start converting all the qt to juce. The main hassle was on the server list. In the end I had to use the url plus the ?directory= but was having issues but was using a older AI at the time so that was probs causing most of them.
I could attempt to fully convert it to juce if you want and go as far as I can go then I reckon you guys could manage to fix anything left.

[ann0see]

My feeling is that it's better to have it as fork and then backport features one by one. Otherwise I doubt we'd be able to handle the review process.

Best case would be some kind of module which wraps all the changes for the VSG and leaves business logic untoucheded - but that's probably not possible with the current code base. You probably already made a big chunk of progress in this direction though.

[AndyMcProducer]

Yeah I think might be better. Don't want it being more work to try find and join old stuff with new stuff to get it all up tidy. But I got it all working, I removed the private that function atm cos it was spamming chat and should have been hidden. :S But refined the auto level to give more clip protection plus if the boost button it more inline with the auto level and if used with auto level will just slightly boost those channels and duck others if needed but subtly so end effect is that the channels you boost whilst in auto level just pop a bit in front of the rest. I forgot about the auto mute function in jamulus and boost kept on muting the channel I tried it on. I have added a basic low latency limiter, shouldn't add nothing isn't for pushing up volume, just for stopping clipping with a soft limit so peeps mixing manually can use it to help protect against sudden loud spikes which could damage speakers and headphones and ears. So it's there more as a backup. Most times it will seem like it's not doing nothing which is what I wanted to achieve, same for auto level, that will increase all volumes into a nice level without clipping and sharp in out volume spikes. I'll build a setup later for it and put a google link. Until I sort the git thing. NYE tomoz so probs be least a week or 2 after that cos busy after the new year but encase I don't see anyone till then HNY everyone. xx

…

________________________________ From: ann0see ***@***.***> Sent: 29 December 2025 17:00 To: jamulussoftware/jamulus ***@***.***> Cc: AndyMc ***@***.***>; Author ***@***.***> Subject: Re: [jamulussoftware/jamulus] Jamulus VST3 (Discussion #3583) My feeling is that it's better to have it as fork and then backport features one by one. Otherwise I doubt we'd be able to handle the review process. — Reply to this email directly, view it on GitHub<#3583 (reply in thread)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AB64GYOK35TW7JXXTXSOZB34EFM4RAVCNFSM6AAAAACP6W45UCVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTKMZWHA2TMNA>. You are receiving this because you authored the thread.Message ID: ***@***.***>

[ann0see]

If I were you, I'd put the source code next to the binary - best case would of course be just pushing it to github. I am quite sure that - as you already use AI - AI can help you how to do this.

[AndyMcProducer]

Just waiting an a few peeps testing it but yeah I'll probs just ask ai . But as far as I can se and tell it's all working ok but yeah needs a few proper tests and hopefully those people will get back to me soon. Been making a vst and standalone to do screen, audio and midi so I can like work in the studio with out paying a heating bill 😄 Yeah could use software about now but would rather somit I can just load up as a vst. Got it all working good but udp and max size issue so Trying some different methods and don't really need more than 10, 15 fps for 4k, long as I can read the words. Was looking into Maybe the filters obs has for scaling down. The nI could just scale down to 1080p. Testing atm gpu encoding and some dif methods, probs end up using ffmpeg for cross platform support. But I'll get it uploaded soon, gonna just do a clean up on the code as probs lots of shit left in their from the AI that is dead non used code. The vst should built to standalone and others like vst2, au, lvl, possibly clap, but im sure their is a clap wrapper about.

…

________________________________ From: ann0see ***@***.***> Sent: 11 January 2026 10:36 To: jamulussoftware/jamulus ***@***.***> Cc: AndyMc ***@***.***>; Author ***@***.***> Subject: Re: [jamulussoftware/jamulus] Jamulus VST3 (Discussion #3583) If I were you, I'd put the source code next to the binary - best case would of course be just pushing it to github. I am quite sure that - as you already use AI - AI can help you how to do this. — Reply to this email directly, view it on GitHub<#3583 (reply in thread)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AB64GYKC5CK6YPRFH7FUWOT4GIRURAVCNFSM6AAAAACP6W45UCVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTKNBWG42DOMA>. You are receiving this because you authored the thread.Message ID: ***@***.***>

[AndyMcProducer]

So I put it out to beta and straight away wouldn't recognise on a remote machine, tested on another pc here and same thing. Dumpbin and dependencies gui just didn't help. I'm thinking is maybe still a QT issue. Probs cos I got QT versions installed on the dev machine. But the method I've used to get it converted I think now will be able to convert to fully juce with no qt as the source is a lib now. Before when I tried to convert it to juce too many things specially the server list had issues probs could have put it up here and you guys would have seen the issues. But now I'm but by bit converting QT stuff to Juce. Read a few articles where they said they used MinGW to compile it which then fixed issue son other pc's. So compiling with that too to see if it helps but still want to move it all to juce and no more QT.

…

________________________________ From: Andy McCance ***@***.***> Sent: 13 January 2026 16:19 To: jamulussoftware/jamulus ***@***.***>; jamulussoftware/jamulus ***@***.***> Cc: Author ***@***.***> Subject: Re: [jamulussoftware/jamulus] Jamulus VST3 (Discussion #3583) Just waiting an a few peeps testing it but yeah I'll probs just ask ai . But as far as I can se and tell it's all working ok but yeah needs a few proper tests and hopefully those people will get back to me soon. Been making a vst and standalone to do screen, audio and midi so I can like work in the studio with out paying a heating bill 😄 Yeah could use software about now but would rather somit I can just load up as a vst. Got it all working good but udp and max size issue so Trying some different methods and don't really need more than 10, 15 fps for 4k, long as I can read the words. Was looking into Maybe the filters obs has for scaling down. The nI could just scale down to 1080p. Testing atm gpu encoding and some dif methods, probs end up using ffmpeg for cross platform support. But I'll get it uploaded soon, gonna just do a clean up on the code as probs lots of shit left in their from the AI that is dead non used code. The vst should built to standalone and others like vst2, au, lvl, possibly clap, but im sure their is a clap wrapper about.

________________________________ From: ann0see ***@***.***> Sent: 11 January 2026 10:36 To: jamulussoftware/jamulus ***@***.***> Cc: AndyMc ***@***.***>; Author ***@***.***> Subject: Re: [jamulussoftware/jamulus] Jamulus VST3 (Discussion #3583) If I were you, I'd put the source code next to the binary - best case would of course be just pushing it to github. I am quite sure that - as you already use AI - AI can help you how to do this. — Reply to this email directly, view it on GitHub<#3583 (reply in thread)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AB64GYKC5CK6YPRFH7FUWOT4GIRURAVCNFSM6AAAAACP6W45UCVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTKNBWG42DOMA>. You are receiving this because you authored the thread.Message ID: ***@***.***>

[iamthechrisb]

@AndyMcProducer this is awesome - great to see this and can't wait to check it out!

[AndyMcProducer]

I wanted to use the Qt QTextBrowse to make sure chat html showed perfect as first I'm gonna put up a slight QT version as I'm then gonna finish converting the rest to juce. Prob with QTextBrowse is it may not work on all DAWs as some are stritch and the chat could not show.
Webview I've not looked to see if that even works on other OS's and it's a bit of a hefty resource to add just for the chat.
Plus I've tried it before and had a lot of hassle trying to get it to use chromium and not IE which could make jamulus chat a security risk too.
So I'm sorta adding a paser to handle the html formatting etc. This also stops html code on servers being used to do unwanted stuff.
I wanted to ass P2P layer on top for things like cams and private messaging etc and even possibly topull audio direct if the round trip was quicker but that was just an idea and not something I was thinking of doing now but the cams and dm's would have been nice but the server will need updating first and them used for the extra options to be available.
The city field atm is appended a V which shows on peeps chans and lets others know who else is on vst, orig this was gonna be used to create
the p2p connections but it's just not safe enough and can't be done using the chat in a hidden way either.

But soon as I have the html issue sorted ill push it up as a fork.

Would it be any advantage if I was able to allow people multiple tracks or maybe add multichannel opus support?
But I wouldn't do multiple channel config etc, it would just be ability to bring a floating mini mixer up on peeps using it then with multiple outs
as vst3 supports that and multiple in's. But I gotta get back to sorting the p2p daw after this. I thought all was done but then qt dependencies on machines that didn't have them installed and it not pulling form the vst dir so did like a cheat method where it bundled them in a dll then the vst3 loads that, sorta like a wrapper but not.
On the juce audiopluginhost.exe when dragging on the vst3 to test it it loads 2 instances so will look at that too before upping the fork.