search

koord-live / KoordASIO

A user-friendly universal ASIO driver for use with any Windows installation. Supports WASAPI (shared and exclusive) and includes config GUI.
https://koord.live
Other
92 stars 5 forks source link

Mono mic devices won't work with Jamulus #2

Closed pgScorpio closed 1 year ago

pgScorpio commented 2 years ago

Jamulus 3.8.1 does not accept mono input devices (almost all USB headsets) with KoordASIO. Same headsets via ASIO4ALL are accepted though.... Shouldn't KoordASIO also always report 2 inputs, even when mono? (Either one silent channel or same input on both channels)

danryu commented 2 years ago

Hi @pgScorpio and sorry for the inactivity here. This is now being addressed.

KoordASIO/FlexASIO actually correctly reports the number of channels, but I think Jamulus itself is being over-zealous in insisting on both stereo input and stereo output. I tested with a USB speaker headset with

Jamulus does indeed fail on trying to load this mono mic as you described, and I traced this behaviour back to this commit from 2007: https://github.com/jamulussoftware/jamulus/commit/74cb27d2bc97860d3163e43a71edd3e994fc790f

but IMO there isn't an obvious reason why this is in place - especially as Jamulus has mono modes of operation(?).

Indeed if you simply comment out the lines: https://github.com/jamulussoftware/jamulus/blob/master/src/sound/asio/sound.cpp#L168-L174 and rebuild Jamulus you will find that it works fine with KoordASIO/FlexASIO :)

Note: I'm not sure how/why ASIO4ALL reports 2 x input channels for a single-channel input device - that may be a quirk or a workaround. I did a basic debug run with Jamulus/Koord-RealTime and ASIO4ALL using that mono USB mic - it did indeed report 2 x channels:

lNumInChan:  2
lNumOutChan:  2

It seems like the better thing to do would be to NOT fail on valid single-channel input devices, rather than rely on this strange behaviour.

One to raise with the Jamulus folks?

pgScorpio commented 2 years ago

Yes, I already reported this issue on the Jamulus github. And I am also working on a better sound implementation in Jamulus myself, but also the Jamulus development is on a low level these days. (I'm already waiting months now for approval reviews.)

danryu commented 2 years ago

Ok yes I see the issue here: https://github.com/jamulussoftware/jamulus/issues/2344

Your changes in the original post look interesting - if you'd be interested, I'd welcome a PR on our (compatible) fork at https://github.com/koord-live/koord-realtime ;) Otherwise I will follow your example changes as given and test myself. Our fork is in fairly heavy development (mostly recently on adding integrated video with the koord.live service) but we have a lot of interesting changes going on in the GUI and on mobile devices. Keep an eye out in the coming weeks for the v2 update.

danryu commented 1 year ago

As this is more focused on Jamulus and its restrictions than the ASIO driver, deferring this issue to https://github.com/koord-live/KoordASIO/issues/4.

@pgScorpio Out of interest, regarding your re-factoring of Jamulus audio codebase that you mentioned earlier: did you consider using the JUCE framework at all, and if not why not?