[Portaudio] ASIO unwanted loopback when channelCount = 1

Bryan Rambo bryanr at bometals.com
Wed Aug 2 05:35:48 EDT 2017


Hi Ross,

Thanks for the insight and advice!

As you suggest below, all four of PA's other Windows APIs just fan-out the
mono output to both L&R; we're only seeing this on PA/ASIO.  Since this is
not something that is immediately recognizable to you as expected behavior
for PA, I'm thinking the soundcard/driver probably is the culprit.

My Lynx Studio (AES3 adapter for PCIe bus) does the funny loopback and we
have two other users reporting the same problem, but don't know what
soundcard they're using.  I will ask them.  I ordered a cheapie Behringer
UMC202-HD for further testing, it should be here today.

Thanks again for the tips,

Bryan

-----Original Message-----
From: Ross Bencina [mailto:rossb-lists at audiomulch.com] 
Sent: Tuesday, August 01, 2017 11:34 PM
To: bryanr at bometals.com; portaudio list
Subject: Re: [Portaudio] ASIO unwanted loopback when channelCount = 1

On 1/08/2017 11:27 PM, Bryan Rambo wrote:
> Is this the expected way for PA to handle channels on the host sound
> device that aren't currently being used for the stream?

Hi Bryan,

For PortAudio as a whole, I don't recall that there's a particular 
expected behavior in this case. (Maybe someone else remembers 
differently?) I'm not even sure whether it makes sense to be consistent 
across all host APIs (for at least two reasons: (1) often we'd sent mono 
to the native API and let it deal with it, (2) for a consumer API it 
would make sense to fan out mono to L,R, whereas for a pro API it 
probably wouldn't).

That said, loop-back on unused channels would rank pretty low on my 
personal list of expected behaviors! so it might be a bug or oversight 
in PA/ASIO. If it were a bug, and someone decided to fix it, because of 
the nature of ASIO as a pro-audio I/O interface, I would expect mono-in, 
mono-out to address only the first channel. On the other hand, I'm 
pretty sure that other native APIs will output a mono buffer to both L 
and R.


The upshot is: if you want a specific behavior, I recommend opening 
2-channel devices as stereo and implementing your own desired behavior.


Regarding a possible bug: Is the behavior present on all 
soundcards/drivers that you've tested, or only some?

Thanks,

Ross.




More information about the Portaudio mailing list