Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[jjkale]

I wanted to expand on WDM issues when measuring with REW from a previous thread.
Although the pops and clicks are a "thing", the latency problems are much more significant and pretty much render measurements useless.
Is it correct that when WDM is not used, the REW test tone does not go through JRiver, but goes directly to the sound card?  This is what it seems like to me.  So impulse measurements in this case (no WDM) are "accurate" or at least "more accurate". 
However, if the WDM driver is in the chain, I get latency readings which are much longer and differ widely on repeat readings.
Frequency measurements seem to be unaffected, but if the timing is off then freq adjustments may not be accurate, especially in the LF.
Full disclosure, I am using a USB input to a MCH I2S PinkFaun Bridge then 3 different DACs.  I know of no other way to make measurements in this configuration.
Thanks very much for any tips or suggestions.

[mwillems]

If you need consistent latency between measurements (and can't manually re-zero the measurements), I might be tempted to try the WASAPI loopback or ASIO line in functions instead of the WDM.  You can just turn those on and leave them open as long as need be, and you could more likely get consistent latency (at least for the duration of that measurement session). 

I seem to recall someone else having the same issue years ago, and the issue was resolved by using WASAPI loopback instead of the WDM driver.

[jjkale]

mw,
Thanks very much.  Is  there a reference for WASAPI loopback?

[mwillems]

mw,
Thanks very much.  Is  there a reference for WASAPI loopback?

I'm not sure what you mean exactly by "reference," but there was a fairly solid wiki page at one point.  You turn it on by using the "Open Live" menu item and choosing WASAPI loopback.   You'll need to set the windows default sound device to something other than your actual sound output; the recommendation is to use a disconnected sound device like your motherboard soundcard or an HDMi output.  Then MC steals the audio from the default device.  It was the method that existed prior to the WDM driver, but it's less flexible because you need a dummy device and have to turn it on manually.  But the turning it on manually is a strength when doing measurements.

[mattkhan]

Use an acoustic timing reference and check the  "Wait for timing reference" box when measuring

[jjkale]

Yes, I have both of these items checked off...acoustic timing reference and check the  "Wait for timing reference" box when measuring.
I believe I am executing the procedure correctly.

[jjkale]

Also, I see in several threads down from here, titled "WDM", there is a reference to WDM "being worked on".  Is there any clarification for this?  Thanks.

[JimH]

Win10 prevents installation of the WDM driver unless you disable "secure boot".   This is due to a recent change in Win10.

[mattkhan]

Yes, I have both of these items checked off...acoustic timing reference and check the  "Wait for timing reference" box when measuring.
I believe I am executing the procedure correctly.

the reported latency will always be relative to the reference channel then which should also be going via jriver and hence the WDM driver latency should not be relevant. If you are getting variable latency in this case then this implies a configuration error. Perhaps you can share an mdat or show some more detail about the setup.

[mwillems]

Use an acoustic timing reference and check the  "Wait for timing reference" box when measuring

Oh, duh, a timing reference; I was not at my best when answering last night 

Thanks for stepping in mattkhan.

That said, I remembered where I'd had the discussion with someone concerning measurements latency, where switching to loopback solved their issue, it's here: https://yabb.jriver.com/interact/index.php/topic,87538.msg619624.html#msg619624

[mattkhan]

fwiw the acoustic reference should avoid this because the reference signal goes via the exact same signal chain end to end so it shouldn't matter whether that end to end latency varies from measurement to measurement or not as long as the relative delay between channels is consistent (which it surely must be otherwise your playback would be all over the place)

[jjkale]

I agree in principle with your explanation above, but in reality this is not the case for me.
When using WDM (Ipc displayed), on back to back to back to back measurements of L and R I get numbers like: 146.67 ft, -257.77 ft-170.42 ft and 153.49 ft.
Clearly these are unusable.
The actual distance to L and R is 10 ft 3 inches and 10 ft 6 inches.

[mattkhan]

Which channel are you using as the reference?

However note that you can't use an acoustic reference for absolute time of flight measurements so not sure what you are expecting here (or what you are attempting to do)

[jjkale]

Left channel.
It is my understanding that for LCRLSRS an acoustic timing reference is an accurate reference to establish timing for 5.0.
Other than distance, what do you advocate?
For subwoofer measurements, agreed, acoustic timing is not.
My objective is to improve room correction particularly below 300 Hz.
Thanks for the help.

[mattkhan]

If your L is the reference then when you measure the L, the impulse should be centred at 0. If this is not happening, your reference cannot be what you think it is.

[jjkale]

Yes, precisely.
And it is not.  The timing is all over the place. 

[mattkhan]

I suggest posting some detail about the config of each device/app in the signal chain

[jjkale]

Sure.
What data would you like to review?

[mattkhan]

I think just start with each step in the signal chain and how it is configured.

[jjkale]

10 TB Synology in a different room, wired connection, 100+ Mbps
Gigabit switch
PinkFaun Streamer, i7700k, 16 GB RAM, only used as a media PC, no extraneous software, OCXO mobo clock
PinkFaun MCH I2S Bridge with OCXO clock; Win 10 Pro
The I2S does not have a driver, it is seen as a device which distributes 7.1 correctly to up to four different DACs via I2S; in my case 3 different DACs for 5.1+1
I2S to
Denafrips Terminator on L and R
Holo Lvl 3 on C and Sub
Matrix X Sabre Pro on Surrounds
analog out to Emotiva XMC-1 PrePro; volume control only
W4S MCH amp
Revel F208s/Revel/PS Audio subs
REW via USB

When attempting WDM measurements, JRiver Media Center is selected as Default in Windows Sounds.  MC 24 is the Output Device for REW and PinkFaun I2S WASAPI is the JR Audio Device.
Ipc is displayed on MC.
Standard settings throughout MC.

FWIW, the timing does not seem accurate when attempted non-WDM also.
Thanks to all for your assistance!

[jjkale]

Also, when I take timing out of the measurements, all the anomalies go away and I get what seems to be an accurate SPL measurement.

[mattkhan]

the detailed kit list is not so important, it's the actual signal chain that matters so

REW -> WDM -> MC -> WASAPI -> speakers
USB Mic  -> REW

right?

what does the IR for a "bad" measurement look like?
when you play the acoustic timing ref, do you hear it play ahead of the measurement as expected with no other extraneous noise?

the behaviour you describe seems consistent with a measurement that "hears" something else each time

[jjkale]

The chain is correct.
Yes, acoustic timing reference sounds correct as well, although there seems to be delay prior to the ''tweet"
It occurs at about 11% on the horizontal scale on the measurement page.

[jjkale]

The impulse response occurs down the timeline in accordance with the numbered timing result in the measurement tab.
If you would like to see 2 x IR plots I can post later.

[mattkhan]

although there seems to be delay prior to the ''tweet". It occurs at about 11% on the horizontal scale on the measurement page.

I'm not sure what this means, is it a longer than expected delay between the acoustic reference signal and the measurement sweep?

btw it occurs to me that the original point of this thread is measuring your system so you could just use this method instead - https://wiki.jriver.com/index.php/Verifying_DSP_Studio#Generate_a_Measurement_Sweep

i.e. save the REW sweep (with timing ref) as a wav, play it back in MC with REW set to wait for the acoustic timing ref

This takes WDM, or any loopback, out of the equation entirely.

If you still get the same problem then you know there is something else in your signal chain causing a problem.