Stereo Tool

Well, at least with the old I/O layer, if one input or output causes a restart, they all restart. So what you're hearing might be perfectly logical. Can you tell me what you're using as inputs and outputs, and what the buffer sizes are?

48khz
Input 1 > WASAPI "Line 1 VAC"
Input 2 > ASIO "miniDSP ASIO driver" @ 64 samples (add)
Normal Out > KS (used to be WASAPI until a few minutes ago) "Multichannel (FW410 Multi)" @ 45ms with no Overrule block size. This is my M-Audio firewire 410 interface.
FM Out > ASIO @ 52ms, 64 samples

One anomaly is that my miniDSP ASIO (USBstreamerB -- optical I/O box) doesn't support anything above 96khz (and certainly outputs at 48khz according to my receiver), even though it's used for the FM Output. I was under the impression FM required 192khz capability, but I'm grateful that it doesn't in this case.

edit: OH! I just realized when messing with the Synchronization with HD Delay, dropping it below -0.08s causes the same sounding anomaly.

At 96 kHz you can use it for FM, but it won't output a pilot or RDS.

I can see why normal output would give you the warning (since ASIO input and output are tightly locked to each other but KS and Wasapi are not, so their buffers won't be as precisely in the center). But I can't explain why you would hear noise on the FM output and not on the normal output. A drop in audio would be expected, but that's just silence.

If the CPU really can't handle the ASIO calllbacks anymore, ASIO output devices will keep repeating what's in the buffer. And since the buffer is very small it will typically give you a buzzing sound. At the same time, no input will be read, so non-ASIO devices will get buffer underruns. So, it kinda sounds like your ASIO callback might be stopping.

- We did change something in the ASIO code some time ago which sets a CPU affinity. I'll check when we did that, and in which builds it's enabled.
Edit: Never mind, that's only in the new I/O layer. So that can't be the problem.

Interesting. Well, after some more frustration with this continuing back to beta 021, I switched back to Enterprise 044 and the buffers no longer make a noise when they reset. Strange because this initially started occurring when I had switched over to Enterprise, and switching back wasn't a fix.

edit: More detail about the noise/anomaly: it starts glitching out, stops for a very short fraction of a second, then continues glitching for the same amount of time as the first time before fixing itself. Except, on the second set of glitching, it seems to fill in the "gaps" with what sounds like the audio that was missing from the first set of glitching.

This may sound stupid, next time when you test it, try to disable network adapter / internet.

_________________
control point
control point2

Hm... So, the behavior makes sense if the ASIO buffer filling lags behind: It's a circular buffer (we keep going around it, as an infinite loop). So what happens is:

Normal situation: Read is just behind write. So, the processing writes to the buffer, the sound card reads from it.
---------r---w------------------

But if the CPU usage goes up and processing happens too late, this is what happens:
----------------------w----r----

At this point, we're reading audio from the previous run through the buffer (but we also clear it on playback). So it plays silence. But the buffer does get filled with new audio.

So if this situation remains for some time (long enough to go through the whole buffer), we come here again:
----------------------w----r----

But this second time, the buffer isn't empty (because it has never been played). So you hear the audio from the previous run.

If r and w are very close to each other and sometimes one lies ahead, sometimes the other, you'll hear a mix of old and new audio.

So what you're describing makes perfect sense if the write starts to lag behind the read. The question is, why does it happen now while it didn't happen before. Any slight reduction in available processing CPU power may cause this to happen or get worse, so I'm still suspecting the Windows updates. For now, I'm going to ignore what you said about Enterprise, because it really doesn'y make sense and it might just be a coincidence...

BETA047 is available.

Did something recetly changed in MB "Dynamic speeds" ?

i see much difference between 9.82 and 9.92 in "max speedup" and "max slowdown" numbers. Now checking is it working same...

_________________
control point
control point2

The display changed. Say your main dyn adjustment setting is set to x/ 2. So it can speed up or slow down by at most a factor 2.

Now you can change the behavior for speedup and for slowdown. So let's say you want to increase the maximum difference by a factor 1.5, to get a maximum speedup or slowdown of a factor 3.

9.82 showed 1.50 (3.00)
9.92 only shows the resulting value, 3.00.

But did audio changed? If i load preset from 9,92 to 9.82 will it be same?

_________________
control point
control point2