Print

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

[GiAnt]

Hi,
I have a pc-based active multichannel audio system. Sound card is a Motu 828 FW mkIII. In a preliminary setting I have verified  that JRiver is highly effective in manage the whole system (8 channels, ASIO driver, FIR based XOs, DRC on VST convolver, etc.). However, in spite of intense investigation I am not able yet to take measurement of individual channels using Holm, REW or SMAART with the proper timing reference (i.e. in the meantime MC reproduce audio through convolvers and PEQ). This hamper me to correctly adjiust the system for delay and phase. I've tried with ASIO multi server playing with the Asio line command in MC, but unsuccesfully.
May I have your help?
Thanking you in advance
Giant

[mattkhan]

Do you mean you want to use loopback while the convolver and other dsp is active?

[GiAnt]

Do you mean you want to use loopback while the convolver and other dsp is active?

Exactly ...

[mattkhan]

Exactly ...

So you want to measure the flight time after the dsp? If so then the easiest way is to copy the signal as it comes out of your dac in your device's mixer app and loop that back (using your mixer app if possible) to the input specified as the timing input in REW. You don't need a timing output, just copy the outbound signal. Make sense?

[GiAnt]

So you want to measure the flight time after the dsp? If so then the easiest way is to copy the signal as it comes out of your dac in your device's mixer app and loop that back (using your mixer app if possible) to the input specified as the timing input in REW. You don't need a timing output, just copy the outbound signal. Make sense?

Frankly, not only flight time, but something like this, with the JRiver's DSP active

[mattkhan]

Frankly, not only flight time, but something like this, with the JRiver's DSP active

I'm not sure what you're showing me there exactly. The reference signal just gives you an absolute delay that is (when used correctly) consistent across measurements hence giving you useful phase/timing info. It doesn't change an individual measurement in any other way.

[GiAnt]

I'm showing you the IRs of my system (right channel, 4 ways) taken with timing reference. With REW this is easy to do using the asio drivers of the sound card. Unfortunately with the JRiver MC WDM driver it seems impossible to make the same kind of measurement. In order to bypass this limit, I've tried ASIO4ALL, ASIOmultiserver, Virtual cable, but unsuccessfully. My request is aimed to solve, possibly, this restriction

[mattkhan]

I'm showing you the IRs of my system (right channel, 4 ways) taken with timing reference. With REW this is easy to do using the asio drivers of the sound card. Unfortunately with the JRiver MC WDM driver it seems impossible to make the same kind of measurement. In order to bypass this limit, I've tried ASIO4ALL, ASIOmultiserver, Virtual cable, but unsuccessfully. My request is aimed to solve, possibly, this restriction

OK so the easiest way to do this is as I described in http://yabb.jriver.com/interact/index.php?topic=102085.msg708266#msg708266

This means configuring REW to use your device's ASIO driver, using jriver's asio line in to accept the signal from REW and also manipulating the channel routing with your mixer app (cuemixfx?).

Do you know how to do this? I can give an example from my setup (which uses an RME device so the mixer bit will look different but same principles apply)

[GiAnt]

Many thanks Mattkhan for your willingness to help me. I 've also tried the ASIO line in method, but I don't know why, in my hands it doesn't work.  Probably it depends by the sound card. I don't think Cuemix fx had the makings of RME software (internal re-routing).This topic has been also discussed here:
http://www.whatsbestforum.com/showthread.php?16922-Subwoofer-2-2-vs-2-1-REW-measurements
By the way, the link you have indicated probably is wrong ...

[mattkhan]

Many thanks Mattkhan for your willingness to help me. I 've also tried the ASIO line in method, but I don't know why, in my hands it doesn't work.  Probably it depends by the sound card. I don't think Cuemix fx had the makings of RME software (internal re-routing).This topic has been also discussed here:
http://www.whatsbestforum.com/showthread.php?16922-Subwoofer-2-2-vs-2-1-REW-measurements
By the way, the link you have indicated probably is wrong ...

That's me (3ll3d00d) replying in that thread

I googled cuemix and it talks about such routing capability, never used it though. If it doesn't then you can use a physical cable instead.

If you post how you used asio line in and what happened then we might be able to debug it.

Do you need absolute time or is relative time sufficient? If the latter then there is another approach that doesn't require loopback you can try.

[GiAnt]

Probably the best thing is to show you my setting with some images ...

1. My audio device is a FireWire Motu 828 Mk III
2. Device setting Motu Audio ASIO
3. DSP studio (DRC filter is in the convolver VST) 2 channels in 7.1 container
4. Config text for convolution
5. Preamplifier is a multichannel Nuforce Mcp-18. Its outputs, in addition to amplifiers, are linked with the sound card (imput) through y patch cables

[GiAnt]

That's me (3ll3d00d) replying in that thread

I googled cuemix and it talks about such routing capability, never used it though. If it doesn't then you can use a physical cable instead.

If you post how you used asio line in and what happened then we might be able to debug it.

Do you need absolute time or is relative time sufficient? If the latter then there is another approach that doesn't require loopback you can try.

Dear Mattkhan,
I use now successfully the line in function of JRiver with the multiclient ASIO of Steinberg. I can indicate as output in both REW and SMAART7 for instance the ADAT output of my MOTU or SPIDF and with the help of an optical or coaxial cable route the signal toward the ADAT or SPIDF input that in turn enter in JRiver and its DSP. I have then looped back the outgoing signals from the different analog outputs of my audio device both digitally and phisically but unfortunately I 'm not able to obtain any useful information, only crazy data. I've tried any possible combination of input and output as reference but unsuccesfully. I've also tried with a second multichannel presonus ASIO device linked to my motu trough firewire, nothing ...
I do not understand where the issue is . May I have your help? I'm going sligthly mad.
Many thanks in advance
GiAnt

[mattkhan]

I 'm not able to obtain any useful information, only crazy data.

crazy means what exactly?

[GiAnt]

Unreliable delay times for instance (excessively long or short), wrong phase tracking, and so on. When I use the Motu analog output bypassing the JRiver and its DSP it's all OK, both REW and SMAART work great, the problem start when I insert in the chain JRiver ...

[mattkhan]

Unreliable delay times for instance (excessively long or short), wrong phase tracking, and so on. When I use the Motu analog output bypassing the JRiver and its DSP it's all OK, both REW and SMAART work great, the problem start when I insert in the chain JRiver ...

This sounds a bit like a feedback loop, hard to say though. You should be able to do this with your mixer app; pick a pair of channels and open jriver asio line in on them, output to some other pair of channels in REW and route them to the corresponding input channels in the router, you then output from jriver to some other pair of channels and route those outputs back to two other inputs in your router, configure REW to use them as the inputs. Fiddle with buffer sizes till works or use a physical cable to help out.

Get one but working at a time and then plug it all together once you know each bit works.

[GiAnt]

Dear Mattkhan, many many thanks for your patience, but something escape me. In order to calculate the flight time (and therefore the phase) of a signal from a single driver to the mic at the listening position I need for a dual channel measurement (FFT). Therefore in the preference of REW I must indicate "use loopaback as timing reference" under preferences > Analysis > impulse response calculation, right? I have tried to put into practice the advices of your previous post (but without using the mix app, CueMix) as follows. Firstly, I have routed "ASIO line in" toward channel 7 of my audio interface and in REW I have indicated as output the channel 5. With a cable I have linked output 5 (REW) with input 7 (JRiver Asio line in).
In this way the signal enter into JRiver and its DSP correctly (convolver with FIR filters for XO, Parametric equalizer) arriving to all left channels outputs (1,3,5,7). Then I have looped back output channel 3 with a cable to input channel 3.
My question is: which channel must I indicate as "Timing Reference output" and "Loopback input" in the REW window "Soundcard"? I have made several attempts, but all unsuccessfully.
May I have a drawing with a graphical diagram of the connections you have conceived? Probably my English is not good enough to understand all you mean.

[GiAnt]

Probably I have found the solution. The issue could be the xlr mic input on the front of the soundcard. I have now used, as mic input, an analog input on the back of my Motu (with the mic inserted in a second, external interface) and the system delay now is right! The mic input probably introduce a delay that alters the dual channel measurements.

[GiAnt]

The system delay now is right, but in any event the issue remains that adding a delay in the Parametric Equalizer of JRiver the phase showed in REW doesn't change ..

[mattkhan]

You need to output 2 channels from REW, say channels 7 and 8, and use them as output and timing respectively. You then loop them back to 2 inputs, say 1 and 2. You open jriver asio line in for 2 channel input on those 2 channels. This comes through jriver as a 2 channel input and is processed as per any stereo input where the L channel is what you want to measure and the R channel is your reference. You now route, in jriver, that R channel to some unused output channel, apply no processing and especially no delays, and loop it back to another, unused, input. You set this input as the timing input in REW. The L channel is processed normally and output to whatever speaker you want to measure.

Hopefully you can see how a mixer app makes this easier.

Make sense?

[mattkhan]

here's a pic

red is the timing reference signal path
black is the measurement signal path
1-8 on the left are inputs and 1-8 on the right are outputs
I assumed a 1 channel speaker measurement for simplicity

[GiAnt]

Great Matt! Your patience and willingness are commendable. I think this thread will be very useful to a lot of people exploring the convenience of a JRiver-based active multi channel audio system. I'm very happy with it and I'm only waiting to solve this issue to abandon any other software or hardware solution. Results I've obtained until now are outstanding. Thanks to your last two post I realized to not be able to prevent the right-reference channel 8 of your schema to enter into the JRiver DSP. Is this probably the last information I need for solve the "puzzle". It could be enough to use a combination of "mute" and "mix" commands in the Parametric Equalizer? "Move" 8 to 6 could work?
cheers

[mattkhan]

Thanks to your last two post I realized to not be able to prevent the right-reference channel 8 of your schema to enter into the JRiver DSP. Is this probably the last information I need for solve the "puzzle". It could be enough to use a combination of "mute" and "mix" commands in the Parametric Equalizer? "Move" 8 to 6 could work?
cheers

it really depends on what you want to measure, just remember that the measured time of flight is just the total round trip for the measurement signal minus the total round trip for the timing signal. For example, perhaps you want to measure latency added by the jriver asio line in or perhaps you want to measure the delay through the convolver, just apply the relevant signal stages to the timing signal and away you go.

In your example, passing it through the DSP block but applying no additional signal processing is probably what you want.

The other thing to bear in mind is that the timing signal should contain similar frequency content to the reference signal if you want the results to make sense (e.g. compared a low passed measurement signal vs a full bandwidth timing signal is unlikely to yield useful data, in this case you may find it easier to piggy back, aka add, the signal onto the same output channel but delay it by a fixed amount & then comparing the measured delta in the impulse vs the expected)

[GiAnt]

Yes, I know the matter (more or less) . My specific aim now is to properly align the subs to the main speakers. Both frequency response and phase of the single drivers has been previously linearized with rephase (phase linear FIR filters) but at the listening position the delay of subs need to be considered in order to have all the impulses correctly aligned for a good step response. On this subject have been written rivers of ink and the audibility of phase at very low frequencies is still debated. By the way, looking the FR of the subs with the polarity inverted at the XO point (80 Hz) I've already seen that this delay is about 11 ms, but I would like to confirm this data with softwares like smaart, easera or REW that allow to display the phase and the impulse thanks to the FFT (dual channel measurement). I'm a relatively recent user of JRiver and I still need to know its exact mode of operation.

[GiAnt]

Dear Mattkhan, using ADAT (that transport 8 channels) as input instead analog 7 of you example it could be detrimental? Finally, to route channel 8 directly toward output 6 without passing through JRiver DSP is it right to use the mix commands of PE2?
Have a great day!
GiAnt

[Solderman]

I am quite new to REW. Like OP, I too am using (up to) 8 channel ADAT with JRiver. In my case for three (soon to be four) stereo channels of digital active crossover. I like REW but am vexed hexed and perplexed by trying to use the loopback and/or other precise timing features so generally don't

Might this work? Generally with analog audio, it is ok for one output to feed two inputs. There may be a drop in signal level depending upon the loads the inputs present (I am using XLR; not sure). In telecomm this is called a bridged connection, used to monitor an active line without loading down the signal level. I don't see why this wouldn't work with a DAC's output. It would require a custom made cable or some kind of splitter. You would just tap into the line you want to measure (say output #4) and route the input to (say) input #2 and tell REW that is the timing reference.

But will REW work like that (with the identical signal being looped back?) Or does it expect a unique signal on the reference input?

[mattkhan]

using ADAT (that transport 8 channels) as input instead analog 7 of you example it could be detrimental? Finally, to route channel 8 directly toward output 6 without passing through JRiver DSP is it right to use the mix commands of PE2?

you can use some other device via ADAT, just need to make sure they sync to the same clock & yes you can use the mix commands in DSP.

But will REW work like that (with the identical signal being looped back?) Or does it expect a unique signal on the reference input?

yes that will work

[GiAnt]

I'm grateful to you for the useful information provided in response to my question. However in spite i've tried all the possible combination offered by DSP module (expexially MIX) I still not able to prevent the reference signal (n.8 of the Mattkhan's example) to enter in the DSP. I would be very grateful to everybody helping me to solve this latter obstacle. My system is configured as 2 channel in 7.1 container and XO are FIR filters indicated in a config text file inside convolver.
Cheers

[mattkhan]

I don't follow. It will be subject to signal processing in JRiver, you have to configure that to pass just that channel through unmolested.

[GiAnt]

I understand but I'm unable to do it. The level of my JRiver knowledge is clearly lower that i assumed ....

[GiAnt]

I don't follow. It will be subject to signal processing in JRiver, you have to configure that to pass just that channel through unmolested.

and therefore simply not use "adjust frequency" "delay" etc for that channel? Convolution with XO is admitted?

[mattkhan]

and therefore simply not use "adjust frequency" "delay" etc for that channel? Convolution with XO is admitted?

yes and maybe (a high passed signal can be ok, low passed is not)

If we talk in concrete terms of what you want to measure then perhaps that will be easier rather than in slightly abstract terms of what it is possible to achieve.

so what exactly do you want to measure? I see you have a stereo 4 way cfg file though I also see you have convolverVST in there as well (not sure what that is for). Do you want to measure one speaker as a whole or individual drivers within a single speaker? or something else?

[GiAnt]

I want measure the single drivers within the respective speakers. The aim, as above mentioned, is to align the different impulses (namely applying to each the right delay) in order to have an "in phase" response at the listening position. Practically matching the phases like shown, here https://www.youtube.com/watch?v=JdUUKmaxZ14. The hope is to have a better sounding system.
The convolver VST is used to load the stereo Wav containing the DRC filter chreated with DRC interface.

[mattkhan]

I want measure the single drivers within the respective speakers. The aim, as above mentioned, is to align the different impulses (namely applying to each the right delay) in order to have an "in phase" response at the listening position. Practically matching the phases like shown, here https://www.youtube.com/watch?v=JdUUKmaxZ14. The hope is to have a better sounding system.
The convolver VST is used to load the stereo Wav containing the DRC filter chreated with DRC interface.

OK I see. You are correcting the input signal with your current config, have you verified the correction is good after applying the various filters in your crossover?

Anyway you have a few options on how to do this. What are the corner frequencies for the filters in your 4 way setup?

A simple approach is to use REW's acoustic timing reference feature which removes the requirement for a physical loopback, you would choose your R speaker as the acoustic timing reference and just measure each driver individually (by muting the other channels of the L speaker in dsp studio). The downside is that it may get the timing wrong for the woofer as it will be attempting to align a lower/slower IR to a HF IR which can be tricky.

Another approach is to repeat the above but with a physical loopback which will give you an absolute time. You'd probably be able to just use the tweeter signal from the R channel for this so you'd want to route that signal *after* it's been through both convolution stages back to REW as your timing reference. You'd then take 4 measurements for each individual driver by muting the other channels in a PEQ block as per the acoustic timing reference approach.

In each case I would also take some combined measurements of each neighbouring way in your speaker (W+M, M+T etc) and then another one of all 4 combined. This will enable you to verify the accuracy of your data because you should be able to sum those measurements (A+B option) using trace arithmetic in REW and see a close match between the calculated result and the measured result.

You can then adjust timing of the various IRs in REW to explore the resulting phase alignment.

[GiAnt]

Ok. Thanks a lot for sharing your experience. I will do as you suggested. In the meanwhile I show you the actual FR of my system at the listening position.
Cheers