Print

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

[6233638]

I was wondering if anyone here had advice or recommendations about room correction.
I don't have a particularly high-end speaker setup right now, as I have mostly been focused on headphones the last few years, but I'm now looking to move in that direction.
 
Before buying better speakers though (I have a pair of OK bookshelf speakers right now) I was wanting to set up some room correction first and see how good I could get those to sound. After all, higher-end speakers should benefit from room correction just as much as the ones I have now.
I've been experimenting with the demo of Audiolense as it lets you take measurements and apply the corrections to a 90-second music sample, and it's definitely making an improvement.
 
 
Right now my setup is all a bit thrown together though. I'm using my old Radioshack 33-4050 SPL meter with a generic calibration file hooked up to the line-in of my PC's on-board sound, and because I wasn't able to find any any long audio cables, I hooked up my Benchmark DAC2 through a long USB extension and used that as the source. (latency was too high via Airplay)
 
So I have some idea of what room correction is going to do for me, but I'm a bit unsure about how to proceed now.
 
 
For now at least, I'm only interested in stereo (2.0) correction, but I will probably want to correct the audio in multiple rooms.
I think this rules out Dirac as I believe they use a driver to process the sound locally on the PC, rather than offering a VST plugin or working with Media Center's Convolver.
Unfortunately, that was probably going to be my first choice as I've heard the Dirac software in use before, and it was very impressive.
As far as I know, both Acourate and Audiolense should work via Media Center's Convolver, and I'm not sure if there are other options that do as well.
 
At first, Audiolense appears to be the cheaper option at $225 for the 2.0 version, but it seems that the $530 XO version actually includes features which would still be relevant to a stereo setup?
On the other hand, there's only one version of Acourate and it's $390, which keeps things simple.
 
It seems like Acourate is the more advanced package, and comes highly recommended, but it's a lot of money when there are no functional demos for it, and Audiolense offers you three months to return it if you're unhappy with the results.
 
I assume that if I'm already spending that much on the software package, I should also be replacing my SPL meter with a calibrated mic - which seems to be in the $150-400 range.
I'm not quite sure what to do about sound hardware though. I've read that you should be using the same sound device for both playback and recording for time domain corrections to be performed correctly. Is that true? I would probably have to invest in more sound hardware if that's the case. (or use the on-board audio)
 
It also looks like Acourate requires you to use a single ASIO device (no MME/DirectSound support) for both input and output.
It seems like you could probably use something like ASIO4All to use any devices you like with it, but there's probably a reason it does things that way?
I was thinking of buying the XTZ Mic that Dirac recommend as it's only $140 and is a USB device, unlike most of the other calibration mics being sold - but that probably wouldn't work with Acourate (at least not without ASIO4All) and not if you need to be using the same sound device for playback and recording in these packages.
 
Another thing I was wondering about was room position. Both Dirac and Audiolense let you take multiple readings from different positions in the room so that it sounds good everywhere rather than only where the Mic was when taking measurements. Is this something Acourate offers as well? Is this feature all that necessary?
 
I know that I've only mentioned Acourate, Audiolense, and Dirac, but that's because it's all I'm aware of. I've not ruled out anything else if there's other software you would recommend instead.

[mwillems]

I assume that if I'm already spending that much on the software package, I should also be replacing my SPL meter with a calibrated mic - which seems to be in the $150-400 range.

I'd recommend getting a condenser microphone of some kind no matter what software you're using; even an uncalibrated mic will probably get you farther than an SPL meter (depending on the meter).  This is what I use currently (it ships with a measurement software suite that provides some useful utilities): http://www.parts-express.com/dayton-audio-omnimic-v2-precision-measurement-system--390-792.  I used phantom-powered mics for a few years, and after having a USB mic around for a while I'm never switching back if I can help it.

I'm not quite sure what to do about sound hardware though. I've read that you should be using the same sound device for both playback and recording for time domain corrections to be performed correctly. Is that true? I would probably have to invest in more sound hardware if that's the case. (or use the on-board audio)

It also looks like Acourate requires you to use a single ASIO device (no MME/DirectSound support) for both input and output.
It seems like you could probably use something like ASIO4All to use any devices you like with it, but there's probably a reason it does things that way?

Several of the automated correction suites make a similar recommendation.  The reasons I've seen for using the same device for input and output relates to latency matching, and if the software doesn't support multiple interfaces, I wouldn't recommend trying to use ASIO4All to trick it into working (ASIO4All's latency is not predictable in my experience).  That said, it seems like the basic problem would be a trivially easy problem to solve when running logsweeps or other artificial calibration tones.  The software knows what the test tone is "supposed" to look like, how long it's supposed to last, etc.   It seems like it would be simple enough to just look for a loose match in the input, and some measurement software cheerfully does this already.  For example, Holm Impulse (which is freeware) doesn't care what your input and output devices are and still measures time-domain responses successfully, and it looks like Dirac has a way to do it as well.  

Another thing I was wondering about was room position. Both Dirac and Audiolense let you take multiple readings from different positions in the room so that it sounds good everywhere rather than only where the Mic was when taking measurements. Is this something Acourate offers as well? Is this feature all that necessary?

The feature is pretty important for an automated system.  

In the low frequency range there are a lot of issues that multiple measurements will help address, but a good example is Room Modes, which affect the perceived loudness of sound differently in different locations in the room: see http://en.wikipedia.org/wiki/Room_modes  and http://en.wikipedia.org/wiki/Loudspeaker_measurement#Room_measurements.  The nature of Room Modes is that different frequencies will seem to be amplified or nulled by different amounts in different places in the room.  So if your measuring position is in a 200Hz null, an automated tool might try (possibly vainly) to fill that null.  The result is that everywhere else in the room 200Hz will sound pretty loud, and the spectral balance will be goofed up.  You can see how that would play out across different fact patterns.  Taking into account multiple measurements will help to distinguish whether the frequency response non-linearity is a function of a specific position in the room, or whether it originates with your speakers (or with the listening area as a whole).

At high frequencies, the wavelength of sound becomes very small in relationship to us.  For example, sound at 3KHz has a wavelength of about 12cm and 8KHz has a wavelength about 4cm. Performing phase correction on sound that localized based on one measurement means that even folks sitting right next to you may have a very different experience of the sound.  Some folks argue that attempting to phase correct high-frequencies is a generically bad idea, but if you take multiple measurements you might discover a necessary common correction (i.e. a phase shift present most places in the room).  

I know that I've only mentioned Acourate, Audiolense, and Dirac, but that's because it's all I'm aware of. I've not ruled out anything else if there's other software you would recommend instead.

I'd recommend you have a look at REW http://www.hometheatershack.com/roomeq/ It's not as powerful or flexible as the three you mentioned, but it has the advantage of being completely free, and might give you an opportunity to get your feet wet with measurement and room correction.

[6233638]

I'd recommend getting a condenser microphone of some kind no matter what software you're using; even an uncalibrated mic will probably get you farther than an SPL meter (depending on the meter).  This is what I use currently (it ships with a measurement software suite that provides some useful utilities): http://www.parts-express.com/dayton-audio-omnimic-v2-precision-measurement-system--390-792.  I used phantom-powered mics for a few years, and after having a USB mic around for a while I'm never switching back if I can help it.

If you're in a pro environment and already own a Mic amplifier, it seems like a phantom powered Mic might be easier, but for this type of use, USB definitely seems the way to go as long as the device doesn't require specific drivers to work.
That looks pretty nice, and I like that it comes in a custom case - it's going on the list of mics I'm considering.

It's funny, I have spent so much on display calibration hardware, but for some reason I have difficulty spending a lot of money on audio calibration.

EDIT: It looks like the meter I have is a condenser mic. It's obviously not going to be on par with something like that of course.

Several of the automated correction suites make a similar recommendation.  The reasons I've seen for using the same device for input and output relates to latency matching, and if the software doesn't support multiple interfaces, I wouldn't recommend trying to use ASIO4All to trick it into working (ASIO4All's latency is not predictable in my experience).  That said, it seems like the basic problem would be a trivially easy problem to solve when running logsweeps or other artificial calibration tones.  The software knows what the test tone is "supposed" to look like, how long it's supposed to last, etc.   It seems like it would be simple enough to just look for a loose match in the input, and some measurement software cheerfully does this already.  For example, Holm Impulse (which is freeware) doesn't care what your input and output devices are and still measures time-domain responses successfully.

You're right - I did some testing with my DAC hooked up to the line-in on the PC, and latency was jumping around from 40-60ms, to 300-400ms every few seconds when it was running through ASIO4All as a single device.
 
I agree that latency should not be affecting the measurements, but it definitely caused problems with some software when I was using Airplay.
 
With Acourate, you cannot use separate input/output devices, you can only select a single ASIO device and use its inputs/outputs. If Acourate is the software to get, I suppose I can pick up a reasonably cheap USB audio interface for those purposes.
I have been wanting to get a nice desktop mic at some point down the line, so perhaps something like a Focusrite Scarlett 2i2 would be a good purchase. (or complete overkill? I'm not sure)

The feature is pretty important for an automated system.  

In the low frequency range there are a lot of issues that multiple measurements will help address, but a good example is Room Modes, which affect the perceived loudness of sound differently in different locations in the room: see http://en.wikipedia.org/wiki/Room_modes  and http://en.wikipedia.org/wiki/Loudspeaker_measurement#Room_measurements.  The nature of Room Modes is that different frequencies will seem to be amplified or nulled by different amounts in different places in the room.  So if your measuring position is in a 200Hz null, an automated tool might try (possibly vainly) to fill that null.  The result is that everywhere else in the room 200Hz will sound pretty loud, and the spectral balance will be goofed up.  You can see how that would play out across different fact patterns.  Taking into account multiple measurements will help to distinguish whether the frequency response non-linearity is a function of a specific position in the room, or whether it originates with your speakers (or with the listening area as a whole).

At high frequencies, the wavelength of sound becomes very small in relationship to us.  For example, sound at 3KHz has a wavelength of about 12cm and 8KHz has a wavelength about 4cm. Performing phase correction on sound that localized based on one measurement means that even folks sitting right next to you may have a very different experience of the sound.  Some folks argue that attempting to phase correct high-frequencies is a generically bad idea, but if you take multiple measurements you might discover a necessary common correction (i.e. a phase shift present most places in the room).  

I'm not sure if Acourate can do this or not - I just noticed that Audiolense and Dirac made it clear that they did use this in their corrections.

I'd recommend you have a look at REW http://www.hometheatershack.com/roomeq/ It's not as powerful or flexible as the three you mentioned, but it has the advantage of being completely free, and might give you an opportunity to get your feet wet with measurement and room correction.

You know, I did give REW a shot last night, but I couldn't figure out how to create correction filters with it, though that was before I switched from using Airplay to my Benchmark DAC, so perhaps that was the problem. (latency was causing all sorts of problems in REW)
 
I've only ever used REW for general measurements and manually doing some EQ or adjusting the gain/crossover on a sub before.

[mwillems]

If you're in a pro environment and already own a Mic amplifier, it seems like a phantom powered Mic might be easier, but for this type of use, USB definitely seems the way to go as long as the device doesn't require specific drivers to work.
That looks pretty nice, and I like that it comes in a custom case - it's going on the list of mics I'm considering.

It's funny, I have spent so much on display calibration hardware, but for some reason I have difficulty spending a lot of money on audio calibration.

EDIT: It looks like the meter I have is a condenser mic. It's obviously not going to be on par with something like that of course.

Well that's good news.  If you've already got a condenser mic, then you could put off the mic purchase until you've settled on a software package/audio interface.

You're right - I did some testing with my DAC hooked up to the line-in on the PC, and latency was jumping around from 40-60ms, to 300-400ms every few seconds when it was running through ASIO4All as a single device.
 
I agree that latency should not be affecting the measurements, but it definitely caused problems with some software when I was using Airplay.

Oh, don't get me wrong, I'm sure it is a problem for some software; I just noted that the problem should be solvable, and some software is insensitive to that issue or includes a solution for it.  Software that expects you to use the same device will probably have problems if you use it with different devices.

With Acourate, you cannot use separate input/output devices, you can only select a single ASIO device and use its inputs/outputs. If Acourate is the software to get, I suppose I can pick up a reasonably cheap USB audio interface for those purposes.
I have been wanting to get a nice desktop mic at some point down the line, so perhaps something like a Focusrite Scarlett 2i2 would be a good purchase. (or complete overkill? I'm not sure)

I think the Scarlett would work fine for these purposes (I think Glynor has one of it's bigger cousins, and Matt just picked up a Scarlett 18i20), you could also use something like the Steinberg UR22.  Before you get too far down that road, though, you had mentioned in another thread that you were thinking about bi-amping, but with a sub.  If that's still on the table you should try and find an interface that has enough outputs for that project too.  The focusrite 2i4 or 6i6 both have four analog outputs, the 18i20 has eight outputs. The Steinberg UR824 I mentioned in the other thread has 8 outputs as well.  All of those interfaces have phantom powered mic inputs as well.  

You know, I did give REW a shot last night, but I couldn't figure out how to create correction filters with it, though that was before I switched from using Airplay to my Benchmark DAC, so perhaps that was the problem. (latency was causing all sorts of problems in REW)
 
I've only ever used REW for general measurements and manually doing some EQ or adjusting the gain/crossover on a sub before.

REW is sensitive to latency in my experience, so that may be the issue.  

As for getting the correction filters out of REW, try file-->export--> export filters as WAV.  That should produce something that can be loaded into the convolution DSP module.  However, REW doesn't (to my knowledge) do FIR filtering, so you can also literally just manually dial in the various specified filters into JRiver's PEQ module one for one.  That has the advantage of skipping all the latency involved in the convolution process, and will work on all sample rates (separate convolution files need to be created for each sample rate).

[6233638]

I think the Scarlett would work fine for these purposes (I think Glynor has one of it's bigger cousins, and Matt just picked up a Scarlett 18i20), you could also use something like the Steinberg UR22.  Before you get too far down that road, though, you had mentioned in another thread that you were thinking about bi-amping, but with a sub.  If that's still on the table you should try and find an interface that has enough outputs for that project too.  The focusrite 2i4 or 6i6 both have four analog outputs, the 18i20 has eight outputs. The Steinberg UR824 I mentioned in the other thread has 8 outputs as well.  All of those interfaces have phantom powered mic inputs as well.

Hmm, that's true. I was of course thinking that if I ended up going down that route I would simply have the avr/dac handle that, but you have the same problem there of the input/output going through separate interfaces. Perhaps this is something which is only an issue with Acourate though. It's definitely not a problem for Dirac when they are recommending a USB mic.

As for getting the correction filters out of REW, try file-->export--> export filters as WAV.  That should produce something that can be loaded into the convolution DSP module.  However, REW doesn't (to my knowledge) do FIR filtering, so you can also literally just manually dial in the various specified filters into JRiver's PEQ module one for one.  That has the advantage of skipping all the latency involved in the convolution process, and will work on all sample rates (separate convolution files need to be created for each sample rate).

Ah, I see - so REW does not have a one-button option where you select your target curve and it just automatically generates a file to use with convolution? (which Audiolense does)

[mwillems]

Ah, I see - so REW does not have a one-button option where you select your target curve and it just automatically generates a file to use with convolution? (which Audiolense does)

It does have automatic filter generation and a one button export, but that's not necessarily the best way to do it.  If you use the "match response to target" option in filter tasks in EQ it will automatically generate a series of filters for you.  If you use File--> export-->export filters as WAV, it will give you a convolution file with all the filters included.

But all convolution files are limited to a single sampling rate, so any convolution-based solution requires you to generate multiple convolution filters for different sampling rates.  

The reason I suggested entering them manually is that REW doesn't do any filtering that requires convolution to perform (it can all be done with IIR filters).  So while there's a convenience value in just exporting it and using convolution, you then have to put up with the added latency that using convolution necessarily entails (which can make a/v sync more challenging).  Moving the filters over manually is less convenient, but sidesteps that latency and saves you the step of making the filters for each sample rate.

Does that make sense?

By the way, on the subject of why multiple measurements are a necessity, have a look at slides 101-121 of this presentation for an excellent illustration of why automated room correction based on a single measurement is a bad idea: http://www.hypex.nl/docs/papers/AES123BP.pdf.  The author, Bruno Putzeys, is an amp, DAC, and DSP system designer who has done some pretty groundbreaking work in the field.  

[6233638]

Well in this case, it's only going to be playing 16/44.1 so I don't have to worry about that, but it's something I'll have to keep in mind in the future.
 
I'm not sure that I will have a chance to do anything about it until next weekend now, but I'll give REW a shot.

[6233638]

I had the opportunity to try REW again this morning, and either I'm doing something wrong, or it's not doing a very good job. All the bass is sucked out (even when sitting at the mic position) and there's a whole lot of treble emphasis when using REW to generate EQ filters.
 
When using the demo of Audiolense without any changes in setup (mic position, playback/recording levels etc.) it's a good improvement over how the speakers normally sound in the room, and I haven't tried anything like measuring from multiple positions yet.
It's difficult to properly test with the way I have things set up right now, but it does seem like Audiolense XO is producing a better sound than Audiolense 2.0, even when I'm just trying to correct stereo speakers. (I guess that's the True Time Domain correction at work?)
 
 
I've now got things setup so that I can view the measurements in Audiolense, but I'm just trying to adjust via the Parametric EQ in Media Center as you suggested.
I'm making some progress in flattening the response, but honestly, I don't really know what I'm doing.

The biggest problem is that I have no idea how to visualize the Q parameter. As I understand it, it's the "width" of the filter, but if I put a filter at 90kHz with a Q of 0.1, 1.0, or 10, I have no idea what that's going to cover. Ideally it would tell me that a filter with a Q of 1 is going to cover a 20Hz band. (±10 from the filter point)
 

Here's a quick example of what's happening so far after maybe 5-10 minutes of switching over to using the parametric EQ. (which is obviously not the same as room correction) These measurements are taken from one of the worse locations in the room, but it seems that EQ'ing from there seems to result in good sound from most positions. Obviously there's still a lot of work to be done.

 
I feel like by the time I'm finished, I'll have a list of 100 filters though with the way things are going so far.  
 
EDIT: After spending a while with this, I think I'm just going to leave it until I have a better space, better equipment, and a better idea of which package it is that I want to buy.

[mwillems]

I had the opportunity to try REW again this morning, and either I'm doing something wrong, or it's not doing a very good job. All the bass is sucked out (even when sitting at the mic position) and there's a whole lot of treble emphasis when using REW to generate EQ filters.

It sounds like it's doing something goofy; How does the measurement look when you measure it once you've dialed in the REW filters?

When using the demo of Audiolense without any changes in setup (mic position, playback/recording levels etc.) it's a good improvement over how the speakers normally sound in the room, and I haven't tried anything like measuring from multiple positions yet.
It's difficult to properly test with the way I have things set up right now, but it does seem like Audiolense XO is producing a better sound than Audiolense 2.0, even when I'm just trying to correct stereo speakers. (I guess that's the True Time Domain correction at work?)

Probably so, the XO has a lot more functionality than the 2.0 as I understand it.

The biggest problem is that I have no idea how to visualize the Q parameter. As I understand it, it's the "width" of the filter, but if I put a filter at 90kHz with a Q of 0.1, 1.0, or 10, I have no idea what that's going to cover. Ideally it would tell me that a filter with a Q of 1 is going to cover a 20Hz band. (±10 from the filter point)

1.4 Q is one octave wide, .66 Q is two octaves wide.  Higher is narrower, lower is wider.  A 1 Q filter is about 1.4 octaves.  For example, a filter at 90 Hz with a 1 Q will mostly affect frequencies between about 35Hz and 250 Hz, but a filter at 40 Hz with a 1 Q will mostly affect frequencies between about 16 Hz and  115Hz.  I'd recommend you download some free EQ software that has a GUI to help you visualize.  I use the freeware program RePhase: http://sourceforge.net/projects/rephase/ . It's primary function is to generate custom convolution filters, but it has a nice, clean graphical interface that will let you dial things in and show you what effect your EQ will have on frequency and phase.  I use it as much for its GUI as for its convolution filter generating capabilities.

Here's a quick example of what's happening so far after maybe 5-10 minutes of switching over to using the parametric EQ. (which is obviously not the same as room correction) These measurements are taken from one of the worse locations in the room, but it seems that EQ'ing from there seems to result in good sound from most positions. Obviously there's still a lot of work to be done.

 
I feel like by the time I'm finished, I'll have a list of 100 filters though with the way things are going so far.  

That's a definite improvement, no bones about it.  Are the measurements similarly improved in other locations?  If so, you've already improved your system performance pretty dramatically IMO.

I know what you mean about the filters; they have a tendency to multiply. But you can't EQ every narrow peak and trough (without compromising performance elsewhere in the room), and you'll learn how to consolidate as you go (often one or two filters can take the place of several once you know what the outcome is "supposed" to be).

BTW, your speakers don't happen to be about 2 feet in front of a wall do they?

EDIT: After spending a while with this, I think I'm just going to leave it until I have a better space, better equipment, and a better idea of which package it is that I want to buy.

Well, you've accomplished quite a bit already, but I understand wanting to wait until things are in a more "final" form.  If you do decide to press on, it's the sort of thing that works better a little bit at a time.  Measurement and correction is impossibly fiddly, and during a long test session our ears/stamina of auditory concentration start to wear out and (at least, in my case) so does my patience    

[6233638]

It sounds like it's doing something goofy; How does the measurement look when you measure it once you've dialed in the REW filters?

That was before I installed ASIO Bridge to route the sound through Media Center's DSP engine, so I didn't get a chance to measure it. By the time I set all that up, I had given up on REW and uninstalled Java. (I really don't want to leave Java on my system) Maybe I'll give it another look next weekend, but I don't know that I have the willpower to spend another afternoon doing this.

1.4 Q is one octave wide, .66 Q is two octaves wide.  Higher is narrower, lower is wider.  A 1 Q is about 1.4 octaves.  For example, a filter at 90 Hz with a 1 Q will mostly affect frequencies between about 35Hz and 250 Hz, but a filter at 40 Hz with a 1 Q will mostly affect frequencies between about 16 Hz and  115Hz.  I'd recommend you download some free EQ software that has a GUI to help you visualize.  I use the freeware program RePhase: http://sourceforge.net/projects/rephase/ . It's primary function is to generate custom convolution filters, but it has a nice, clean graphical interface that will let you dial things in and show you what effect your EQ will have on frequency and phase.  I use it as much for its GUI as for its convolution filter generating capabilities.

Thanks, I just saw your write-up on the parametric EQ, so I have a better idea of it now. I suppose the thing I don't understand is why "Q" exists at all. I mean, I know that it makes sense from an engineer's point of view, but why not just tell me the range a filter is going to affect? So if I put a filter at 90Hz with a Q of 1, just tell me that it's going to affect 35-250Hz rather than assume I know what Q means.
 

That's a definite improvement, no bones about it.  Are the measurements similarly improved in other locations?  If so, you've already improved your system performance pretty dramatically IMO.

Yes, it was a similar improvement in most places. I think the problem is that once you start doing very fine-grained adjustments, it becomes a lot more location specific.

What I suppose I ought to do is measure a lot of the possible listening locations in the room, average them, and find a spot which measures close to the average and work off that.
I spent a bit of time after that getting it looking quite a bit flatter still, but being careful not to smooth out the peaks/troughs too much, so that you ended up with a very uneven response across the room. It's sounding good, but not as good as the results from Audiolense XO. (and presumably what Acourate or Dirac could do as well)

Well, you've accomplished quite a bit already, but I understand wanting to wait until things are in a more "final" form.  If you do decide to press on, it's the sort of thing that works better a little bit at a time.  Measurement and correction is impossibly fiddly, and during a long test session our ears/stamina of auditory concentration start to wear out and (at least, in my case) so does my patience  

I only had so much time to spend with it today, and it's really something where spending a lot of time on it feels like it's a task which could just be completely automated... which I suppose is what the products like Audiolense are doing.
That said, I do know from experience that calibrating a display by hand often results in better performance than completely automated calibration, so I suppose it's the same thing there.

[mwillems]

Thanks, I just saw your write-up on the parametric EQ, so I have a better idea of it now. I suppose the thing I don't understand is why "Q" exists at all. I mean, I know that it makes sense from an engineer's point of view, but why not just tell me the range a filter is going to affect? So if I put a filter at 90Hz with a Q of 1, just tell me that it's going to affect 35-250Hz rather than assume I know what Q means.

I agree that it would be easier to read that way for sure.  The filters are emulating analog filters, and Q is one of the mathematical parameters that define the filters in analog.  Also (and this part is a headache) the higher the amplitude of the filter the wider the range it will have a meaningful effect on. The majority of the effect will be in the range I described for any amplitude, but, say, a 10 dB boost will have a very "long tail."  I've attached two example filters from RePhase.  Both are 1Q 90Hz filters, one is 3dB, one is 6dB. 

Yes, it was a similar improvement in most places. I think the problem is that once you start doing very fine-grained adjustments, it becomes a lot more location specific.

Yes.  Most of those small dips and peaks are the result of room echoes and cancellations, and they'll differ everywhere throughout the room.

What I suppose I ought to do is measure a lot of the possible listening locations in the room, average them, and find a spot which measures close to the average and work off that.

That's a good idea. One other tip: when you're sitting back down with this again, try measuring very close to the speaker for an additional data point.  Try at one meter (or slightly closer) with the mic positioned between the two drivers.  Then try measuring with the mic directly in front of the mid-bass element, about 2 or 3 inches away.  The latter measurement won't give you any useful information about the treble region, but will give you a better idea of what your speakers natural bass response is (at that distance the direct output will tend to dominate the room response).  Better would be to conduct those kinds of measurements outside, or, barring that, temporarily move your speakers farther away from the walls of the room they're in for measurement purposes (at least five or six feet if possible).  I'd guess your speakers are about two or two and a half feet from the back wall right now based on the measurements?

Generally, you can safely EQ any peaks or dips that result from the speaker's response, it's the room that's harder to fix in EQ.

I spent a bit of time after that getting it looking quite a bit flatter still, but being careful not to smooth out the peaks/troughs too much, so that you ended up with a very uneven response across the room. It's sounding good, but not as good as the results from Audiolense XO. (and presumably what Acourate or Dirac could do as well)
I only had so much time to spend with it today, and it's really something where spending a lot of time on it feels like it's a task which could just be completely automated... which I suppose is what the products like Audiolense are doing.
That said, I do know from experience that calibrating a display by hand often results in better performance than completely automated calibration, so I suppose it's the same thing there.

Some parts can potentially be automated (calculating the filters, applying them, etc.); other parts can't be automated (learning measurement technique, understanding where the automated tools may overreach, etc.), so you're not wasting your time learning about the process, and it sounds like you're getting some improvements in system performance into the bargain.  

Even with a theoretically perfect fully automated solution there's also always room for tweaking its target for individual taste, your own ears, etc.  For example, many people don't prefer the sound of a flat frequency response, they prefer a gradual roll down across the frequency band from bass to treble (sometimes called a house curve or x-curve).  Alternatively, I think Bob Katz opined on the forums that he likes it flat to 1 or 2 KHz and then gradually attenuating to about 10 dB down at 20KHz.  

Even if you wind up going with Audiolense, there'll probably still be some room for tweaking.

[)p(]

I know that I've only mentioned Acourate, Audiolense, and Dirac, but that's because it's all I'm aware of. I've not ruled out anything else if there's other software you would recommend instead.

There is also this open source drc solution aptly named drc.
http://drc-fir.sourceforge.net

[Mitchco]

I was wondering if anyone here had advice or recommendations about room correction.
 
I know that I've only mentioned Acourate, Audiolense, and Dirac, but that's because it's all I'm aware of. I've not ruled out anything else if there's other software you would recommend instead.

I have tried all of the software mentioned in this thread.  On my system, I have had the best results with Acourate.  I wrote a how to article here: http://www.computeraudiophile.com/content/529-acourate-digital-room-and-loudspeaker-correction-software-walkthrough/

There is a specific target response that offers a perceptually flat frequency response at the listening position, that is covered off in the article.  There seems to be some consensus on this as various people have tried in several different systems and rooms and consistently get the desired result.

I used to use REW, but the issue is that, for room correction, REW does not use frequency dependent windowing (FDW) and therefore the result does not match what our ears hear.  At least that has been my experience.

Another good website is: http://digitalroomcorrection.hk/http___www.digitalroomcorrection.hk_/Welcome.html and this is a nice article: http://www.acourate.com/freedownload/TonyKnightSystemDescription.pdf .

Hope that helps.  Cheers, Mitch

[6233638]

I have tried all of the software mentioned in this thread.  On my system, I have had the best results with Acourate.  I wrote a how to article here: http://www.computeraudiophile.com/content/529-acourate-digital-room-and-loudspeaker-correction-software-walkthrough/
There is a specific target response that offers a perceptually flat frequency response at the listening position, that is covered off in the article.  There seems to be some consensus on this as various people have tried in several different systems and rooms and consistently get the desired result.
I used to use REW, but the issue is that, for room correction, REW does not use frequency dependent windowing (FDW) and therefore the result does not match what our ears hear.  At least that has been my experience.
Another good website is: http://digitalroomcorrection.hk/http___www.digitalroomcorrection.hk_/Welcome.html and this is a nice article: http://www.acourate.com/freedownload/TonyKnightSystemDescription.pdf .

Hope that helps.  Cheers, Mitch

Thank you for the links and recommendation - I read your article yesterday and found it very informative. I also tried using that curve in the Audiolense trial.
It's a shame that Acourate does not have a fully functional demo. You can send in your measurements and a couple of tracks to receive processed tracks, but that's not nearly as convenient as being able to generate them in the demo when there are so many options to explore.
 
As you are a user of Acourate, can you tell me if it lets you take measurements from multiple listening positions when building a correction?
Ideally you would treat the room to even out the response rather than use digital correction, but that's not really an option here.

[Mitchco]

Thank you for the links and recommendation - I read your article yesterday and found it very informative. I also tried using that curve in the Audiolense trial.
It's a shame that Acourate does not have a fully functional demo. You can send in your measurements and a couple of tracks to receive processed tracks, but that's not nearly as convenient as being able to generate them in the demo when there are so many options to explore.
 
As you are a user of Acourate, can you tell me if it lets you take measurements from multiple listening positions when building a correction?
Ideally you would treat the room to even out the response rather than use digital correction, but that's not really an option here.

I mention the target curve as other DRC software I have tried seemed to have different ways of interpreting the curve.  In one, the correction seemed to either overshoot or undershoot the target, no matter what I did.  In Acourate, the first time was the last time.

Wrt to measuring multiple listening positions, I did try that in other DRC software and found, at least in my setup, that taking measurements in multiple listening positions did not produce a better corrected response.  The best results I got was to triangulate on the listening position/speakers and take one measurement and let the speakers natural polar response deal with various other listening positions.  With frequency domain windowing, the window is open in the 100's of milliseconds in the low end and less than a millisecond at the very top.  With a FIR filter length of 65,536 taps and 100's of milliseconds of correction in the low end, the bottom end sounds good in my listening room no matter where I sit.

To answer the question, there is not a ready to go function like that in Acourate as far as I know, but I did not look for it as I don't require it.  But I would not be surprised if it could be done as Acourate is more of a digital audio toolbox with many possibilities.  If really interested, you may want to ask in the Acourate forum: http://groups.yahoo.com/neo/groups/acourate/info

Yes, I still use Acourate and wrote this advanced article.  Hopefully I am not spamming and folks find it useful:
http://www.computeraudiophile.com/content/556-advanced-acourate-digital-xo-time-alignment-driver-linearization-walkthrough/

Cheers, Mitch

[mwillems]

I mention the target curve as other DRC software I have tried seemed to have different ways of interpreting the curve.  In one, the correction seemed to either overshoot or undershoot the target, no matter what I did.  In Acourate, the first time was the last time.

Wrt to measuring multiple listening positions, I did try that in other DRC software and found, at least in my setup, that taking measurements in multiple listening positions did not produce a better corrected response.  The best results I got was to triangulate on the listening position/speakers and take one measurement and let the speakers natural polar response deal with various other listening positions.  With frequency domain windowing, the window is open in the 100's of milliseconds in the low end and less than a millisecond at the very top.  With a FIR filter length of 65,536 taps and 100's of milliseconds of correction in the low end, the bottom end sounds good in my listening room no matter where I sit.

To answer the question, there is not a ready to go function like that in Acourate as far as I know, but I did not look for it as I don't require it.  But I would not be surprised if it could be done as Acourate is more of a digital audio toolbox with many possibilities.  If really interested, you may want to ask in the Acourate forum: http://groups.yahoo.com/neo/groups/acourate/info

Yes, I still use Acourate and wrote this advanced article.  Hopefully I am not spamming and folks find it useful:
http://www.computeraudiophile.com/content/556-advanced-acourate-digital-xo-time-alignment-driver-linearization-walkthrough/

Cheers, Mitch

That really is an impressive article Mitchco.  I read it through a few times last night, and even though I don't use Acourate, I definitely learned a thing or two    As I was reading it, it occurred to me that, while Acourate has some pretty unique functionality, some of the things you're doing for your system with Acourate could potentially be done manually with JRiver and free measurement software (like Holm Impulse, which also offers frequency dependent windowing).  Obviously Acourate can do things that would be much harder (or impossible) to do without proprietary software, but even folks who don't use Acourate could definitely benefit from reading your article and digesting the underlying principles.  There's a lot of good acoustic theory in there.

Which all got me thinking: would there be any interest around here in something similar to Mitchco's article, except focused on crossover and speaker optimization that can be accomplished using only JRiver and freeware?  Obviously it won't get you quite as far, but can get you a good distance.  I say that because I'm getting ready to start work on a pair of bi-amped bookshelf speakers that I plan to finish putting together and tuning over the next month or two.  Because I'll be going through the steps anyway, I could document the measurements/process as I go.  

And I hope I'm not stepping on your toes Mitchco, I just found your article very inspiring

[Mitchco]

That really is an impressive article Mitchco.  I read it through a few times last night, and even though I don't use Acourate, I definitely learned a thing or two    As I was reading it, it occurred to me that, while Acourate has some pretty unique functionality, some of the things you're doing for your system with Acourate could potentially be done manually with JRiver and free measurement software (like Holm Impulse, which also offers frequency dependent windowing).  Obviously Acourate can do things that would be much harder (or impossible) to do without proprietary software, but even folks who don't use Acourate could definitely benefit from reading your article and digesting the underlying principles.  There's a lot of good acoustic theory in there.

Which all got me thinking: would there be any interest around here in something similar to Mitchco's article, except focused on crossover and speaker optimization that can be accomplished using only JRiver and freeware?  Obviously it won't get you quite as far, but can get you a good distance.  I say that because I'm getting ready to start work on a pair of bi-amped bookshelf speakers that I plan to finish putting together and tuning over the next month or two.  Because I'll be going through the steps anyway, I could document the measurements/process as I go.  

And I hope I'm not stepping on your toes Mitchco, I just found your article very inspiring

Thanks and go for it!  I for one would find it interesting and useful as I was going to try Holm Impulse and RePhase, but could not find any docs on how to actually use it.  Many would be grateful for your efforts would be my take.

[mwillems]

Thanks and go for it!  I for one would find it interesting and useful as I was going to try Holm Impulse and RePhase, but could not find any docs on how to actually use it.  Many would be grateful for your efforts would be my take.

I know what you mean; the Holm documentation is pretty lacking, and the discussions over at the "official" threads often shed a little more heat than light on the issues.  There are a few RePhase tutorials out there, but the good ones just happen to be in French 

I'll see if I can't get my ducks in a row over the holidays.

[Arindelle]

My problem is I can't really can't do any "physical" room correction .... it is the worst room imaginable for audio - one wall is glass, different ceiling height from 2.5 meters to almost 8 meters. Floor in stone tiles - no rugs possible. Speaker  placement atrocious. (no choice without jackhammering my fireplace to pieces) So all this is very interesting. Thanks Mitchco

would there be any interest around here in something similar to Mitchco's article, except focused on crossover and speaker optimization that can be accomplished using only JRiver and freeware?

Yes very much so! Can't swing buying acourate, and even with a good guide, the learning curve looks steep. I can't adjust my active analog  Xos really

Really thinking of doing digital active crossovers. And I like "free"! (Would jriver be providing microphones like this at the same price as a their remote ?)

[6233638]

I've spent some time looking over hardware again today, and I'm still not entirely clear on what to get. As far as software is concerned, I think I'm convinced that Acourate is what I want to get now.
 
I don't think there is anywhere here selling calibrated USB mics though - my only option would be to have that Dayton OmniMic shipped over from the US, and I'm not clear on what the shipping charges would be. (I hate that websites require you to register first)
I'm somewhat hesitant to go with a USB mic now though, as I've seen a lot of reports about high noise floors with them (though nothing about the OmniMic) and that they operate at a fixed gain so that measurements have to be very loud to get a good signal level.
 
Alternatively, there are a few calibrated phantom power mics available here, and buying one of them and a Scarlett 2i2 or Steinberg UR22 is likely going to come to around the same price as the OmniMic plus shipping and import fees. I don't know that either the 2i2 or UR22 have ASIO drivers though, so there's a chance I would still have to go through ASIO4All anyway. (Acourate only works with ASIO devices) This would also mean that I'm probably spending as much on the interface as the mic, rather than it all going to the mic.
  
While I am still hesitant, it does seem that using ASIO4All should not be a problem with Acourate though:

Indeed it is the only chance to use two different devices for playback and recording by combining them with Asio4All.
This will work but some attention is required regarding clocks. If two clocks are in game for playback and recording (e.g. USB mic) then they will not fit together by high chance. But it is possible to compensate for this. The Acourate logsweep recorder allows to add Dirac pulses to a sweep. The distance is known. So if the recording shows another distance it is possible to create an inverse logsweep (required for pulse response calculation) with the correct length. Thus the pulses will be correct

But I'm unsure about whether this is something that will have to be done for each measurement, which would be a major nuisance (latency in ASIO4All seems variable) or if it's just something that has to be done once.
With display calibration, some of my hardware requires a dark calibration every 10-20 minutes, and that sort of thing quickly becomes very frustrating.
 
Should I maybe look into battery-powered Mics, or using external amplifiers, and simply buying a PCIe sound card that has an ASIO driver?
I'd really prefer to be using a USB device rather than PCIe though - that way I'm future-proofed if I ever wanted to use a laptop for measurements.

[mwillems]

I've spent some time looking over hardware again today, and I'm still not entirely clear on what to get. As far as software is concerned, I think I'm convinced that Acourate is what I want to get now.
 
I don't think there is anywhere here selling calibrated USB mics though - my only option would be to have that Dayton OmniMic shipped over from the US, and I'm not clear on what the shipping charges would be. (I hate that websites require you to register first)
I'm somewhat hesitant to go with a USB mic now though, as I've seen a lot of reports about high noise floors with them (though nothing about the OmniMic) and that they operate at a fixed gain so that measurements have to be very loud to get a good signal level.

The noisefloor on mine is fairly low. I haven't measured it successfully, but I haven't run up against it in my measurements (the noise floor in my room is higher than the noise floor on the mic).  It is fixed gain, but I can tell you that I have to turn the windows audio input levels way down to avoid overloading it.  Low sensitivity is not the issue with the Omnimic.  But it isn't the cheapest solution, and if you're definitely planning to do Acourate, you might want to go phantom powered with an external interface.

Alternatively, there are a few calibrated phantom power mics available here, and buying one of them and a Scarlett 2i2 or Steinberg UR22 is likely going to come to around the same price as the OmniMic plus shipping and import fees. I don't know that either the 2i2 or UR22 have ASIO drivers though, so there's a chance I would still have to go through ASIO4All anyway. (Acourate only works with ASIO devices) This would also mean that I'm probably spending as much on the interface as the mic, rather than it all going to the mic.

The UR22 definitely has an ASIO driver (it's discussed in the manual http://download.steinberg.net/downloads_hardware/UR22/UR22_documentation/UR22_OperationManual_en.pdf).  Steinberg actually maintains the ASIO specification, so I'd be shocked if it didn't.  I haven't used the UR22, but the UR824 ASIO driver has been completely frictionless for me.  Maybe somebody with a Scarlett can comment as well?  

[mojave]

Any pro audio device will have an ASIO driver. I own the Steinberg UR824, but I also have owned several Tascam products and currently have a Tascam US-366. It is 4 channels so can be used as a preamp and for a 2 channel system with 1 or 2 subwoofers. If you only need a preamp or 2 channels, then I would recommend the Tascam US-122MKII. I've used one quite a bit with REW.

I own a Behringer ECM8000 microphone (same as sold by Acourate) but my current measurement microphone is an iSEMcon EMM-7101-CHTB. It is much more accurate, lower noise floor, and flat from 5 Hz to 20 kHz. iSEMcon is in Germany and Audiolense sells their MP-1r-KIT measurement kit which includes a preamp. Unless you need to measure below 20 Hz or at high SPL levels, then this would work great. You can buy direct from iSEMcon, too.

If you do get something like the Behringer ECM8000, then it needs to be professionally calibrated. You can load the calibration file in Audiolense or REW. I'm not sure about Acourate. iSEMcon individually calibrates all their mics and includes a calibration file.

I've also used an Omnimic. It is an excellent solution and extremely easy to use. It is very easy to take to someone else's house and you can take a measurement in just a few seconds since it uses playback from a CD. The other programs generate their own signal which makes it a lot more difficult to measure on another system than your own.

I use Audiolense XO and really like it. The system setup, measurement workflow, and filter creation is very nicely laid out in my opinion and makes measuring a 7.1 system with multiple subwoofers extremely easy. You can use Audiolense for free to take measurements, export them, and then import them into REW for filter generation. You get a better measurement than REW for the reason's Mitcho provided, but can still use REW for analysis and filter generation. This is a great free solution, IMO.

I take my measurements at 48 kHz and then generate 48 kHz filters with Audiolense. JRiver will automatically resample the filter to match the current output. I have found this simpler than creating multiple filters for each sample rate. Also, my 48 kHz measurement provides the best convolution results with the lowest noise floor.

Here is how Dirac Live handles different input/output devices:

... as far as I know other DRCs require to connect the mic to the same usb port as the output device to make sure they share the same clock source (so a USB mic on one port and a DAC on another USB port is a problem). But you may have noticed that when you do each stereo measurement with Dirac Live three sweeps are played instead of two as expected...this way you can have completely different clock sources on the playback and recording device because we compensate for the timing problems. This is why we play three sweeps when we do a stereo measurement, first left, then right, then left again. When we do like this we can calculate the clock drift between the playback and recording device and compensate for it.

I started a thread on Audiolense's forum 11/20 asking for similar support. However, if their is variable clock drift then this method still doesn't work.

Mitcho - thanks for your article. I have everything I need to go fully active with my main speakers, but still haven't taken that step.

[mwillems]

You can use Audiolense for free to take measurements, export them, and then import them into REW for filter generation. You get a better measurement than REW for the reason's Mitcho provided, but can still use REW for analysis and filter generation. This is a great free solution, IMO.

That's a great point, I've been doing the same thing with Holm for a while (using it as a measurng tool and then exporting to REW) because Holm allows for frequency dependent windowing and/or gating.  It hadn't occurred to me to use Audiolense's measurement suite and do the same thing.  Thanks for the tip  

I take my measurements at 48 kHz and then generate 48 kHz filters with Audiolense. JRiver will automatically resample the filter to match the current output. I have found this simpler than creating multiple filters for each sample rate.

I'll confess that I never got that feature working right, personally.  I know it works for folks, but I never got multiple sample rates working correctly with the same filter (they always sounded mighty odd).  Maybe I'll take another run at it and see if I have better luck now.

[6233638]

Thanks again everyone. At some point, I had been thinking about purchasing an ADC (probably a Benchmark ADC to go along with my DAC) but my requirements for recording are really not going to be that high end.
It looks like one of those Steinberg interfaces would be perfect though, as it will serve as an input/output device for Acourate without any latency problems, and be useful for recording purposes if I buy another mic. (rather than a measurement mic)
 
The UR824 looks like a great piece of kit, but it's a lot more than I'm wanting to spend right now.
I could stretch to the UR44 which I'd prefer to have due to its wider selection of inputs/outputs and the half-rack size (same as Benchmark) but it requires external power and won't be available until Q1 2014.
The UR22 seems like it would keep things as simple as possible, being completely bus powered. Just a shame it's a 1/3 rack width device.
 
Of course at some point down the line, I may want to move to a multichannel setup to use a digital crossover, but realistically that's not going to happen any time soon. Perhaps then I will be looking into the UR824 or similar units.
 
Now I just need to see what mics are available to buy here with individual calibrations. The Behringer ECM8000 is easy enough to get, but there might be some better options that don't cost a lot more.

I'll confess that I never got that feature working right, personally.  I know it works for folks, but I never got multiple sample rates working correctly with the same filter (they always sounded mighty odd).  Maybe I'll take another run at it and see if I have better luck now.

Is it really that difficult to set up multiple sample rates anyway? It seemed like it would be easy enough to leave everything as it is, switch the sample rate and take another set of measurements.

[mwillems]

Now I just need to see what mics are available to buy here with individual calibrations. The Behringer ECM8000 is easy enough to get, but there might be some better options that don't cost a lot more.

I had a calibrated ECM-8000 before I had an Omnimic.  It will work, but it does have a few quirks to get used to (positioning is more important than it should be with a condenser mic, very low frequency sensitivity isn't the best), but it was perfectly satisfactory for my measuring purposes.  I only switched to get a slightly less "positional" mic and for the convenience of USB.

Is it really that difficult to set up multiple sample rates anyway? It seemed like it would be easy enough to leave everything as it is, switch the sample rate and take another set of measurements.

No, once you figure out the naming convention/config file writing it's relatively easy.  Just repetitive.  Although some tools (like REW) don't support all sample rates as outputs, so there's that.  And it may not even turn out to be an issue for you, anyway. 

[6233638]

I had a calibrated ECM-8000 before I had an Omnimic.  It will work, but it does have a few quirks to get used to (positioning is more important than it should be with a condenser mic, very low frequency sensitivity isn't the best), but it was perfectly satisfactory for my measuring purposes.  I only switched to get a slightly less "positional" mic and for the convenience of USB.

It looks like I will probably buy a mic from iSEMcon, as they sell calibrated Behringers and their own mics: http://www.acousticsshop.isemcon.com/index.php?cPath=21_23
One of theirs with a UR22 looks like it will cost me about the same as getting an OmniMic shipped over.
While there'll be a few more pieces involved, with the UR22 being bus-powered it should be as easy to use as a dedicated USB mic, and a bit more flexible as I can repurpose the UR22 later.

No, once you figure out the naming convention/config file writing it's relatively easy.  Just repetitive.  Although some tools (like REW) don't support all sample rates as outputs, so there's that.  And it may not even turn out to be an issue for you, anyway.

Well in this case, the speakers will be hooked up to an AirPort Express which only outputs 44.1kHz, so it won't be an issue, but in the future multiple sample rates will matter once I get speakers set up elsewhere.

[mojave]

Is it really that difficult to set up multiple sample rates anyway? It seemed like it would be easy enough to leave everything as it is, switch the sample rate and take another set of measurements.

You have the following options:

1.  Use 1 measurement and have the measurement software create filters for each sample rate. JRiver will automatically switch filters if named correctly.
2.  Use 1 measurement, one filter, and have JRiver resample the filter.
3.  Use a measurement for each sample rate and a filter for each sample rate.

I choose #2 because I think it sounds better than #1 when I've done comparisons, it is easier to manage all my filters, the hardware and microphone noise floor is lower at 48 kHz than higher sample rates, and I don't have to go through a long manual process of finding the best filter for the measurement when doing multiple sample rates. So far, each measurement is slightly different and seems to require tweaking for the best sound. When doing 6 measurements and 6 filters for a 7.2 surround system it can take a while. Also, I do different filters for different media types so that would be 24-30 filters I would need to create. Now I only need 4-5.

#3 is also dependent on your hardware. You may have 192 kHz output, but the preamp can only record at 96 kHz, for example.

I forgot to mention earlier than I currently just do my filters with a single measurement. I plan to do some multi-seat measurements sometime, but haven't gotten around to it.

[dallasjustice]

I also use the TASCAM unit per Mojave's recommendation. It's great.

I have used all three software and prefer DIRAC LIVE. I can't say I am expert level like Mojave's is to audiolense or Mitch is to both audiolense and acourate. I can say that right "out of the gate" I get much better sound with DIRAC. I really think Bruno is right that multi measurements need to be done (at least in my room). JJ Johnston also did an AES paper on the same topic and he concluded that a tetrahedron shape works well for DSP measuring.

Michael,
I know DIRAC claims to able to account for clock drift, but not with my crazy hardware. :-)  I still use a synchronous measurement. Now I am using my RME card, with a grace m101 and earthworks mic.

Any pro audio device will have an ASIO driver. I own the Steinberg UR824, but I also have owned several Tascam products and currently have a Tascam US-366. It is 4 channels so can be used as a preamp and for a 2 channel system with 1 or 2 subwoofers. If you only need a preamp or 2 channels, then I would recommend the Tascam . . .
I started a thread on Audiolense's forum 11/20 asking for similar support.

[realysm42]

I'm not sure if I've read into this correctly (lots to take in) but I'm concerned about the combination of hardware required to make this stuff work properly.

I've bought this measurement kit: MP-1r-KIT Acoustical measurement kit

My worry is all the mention of latency and using the same input/output device.

My pc has on board sound, but I have an outboard dac (same as most here, I'd imagine) which I feed a signal via asynchronous usb.

Will this be a problem with Dirac/Acourate?

*I know that I plug the mic into the audio input on my pc's motherboard,
*get the software to make it's noises and take the measurements
*create the corrections with whichever program I choose
*and apply the convolution to JRiver

Isthat right?

[dallasjustice]

I am told it won't matter with DIRAC. But I use a very unusual DAC so it matters with my DAC.  I would say it probably doesn't matter with your DAC. With DIRAC you can use a different input and output device. That's a good measurement kit. If I were you, I wouldn't use the mobo mic. I have no experience there but I doubt you will be happy with it. I would get a single device to go out and in like the TASCAM models previously mentioned. That's a much cleaner solution. All of the software mentioned are pretty straightforward to use. However , the filter generated can only be as good as the measurement. GIGO.

I'm not sure if I've read into this correctly (lots to take in) but I'm concerned about the combination of hardware required to make this stuff work properly.

I've bought this measurement kit: MP-1r-KIT Acoustical measurement kit

My worry is all the mention of latency and using the same input/output device.

My pc has on board sound, but I have an outboard dac (same as most here, I'd imagine) which I feed a signal via asynchronous usb.

Will this be a problem with Dirac/Acourate?

*I know that I plug the mic into the audio input on my pc's motherboard,
*get the software to make it's noises and take the measurements
*create the corrections with whichever program I choose
*and apply the convolution to JRiver

Isthat right?

[natehansen66]

Now I just need to see what mics are available to buy here with individual calibrations..

I'd suggest getting a mic from Herb at Cross Spectrum Labs. He calibrates Dayton mics to his professional reference. You get multiple calibration files (for using the mic at different angles) on a thumb drive and printouts of the mic's response and calibration. Good prices and I think he offers the usb mic as well.

[mwillems]

I'm not sure if I've read into this correctly (lots to take in) but I'm concerned about the combination of hardware required to make this stuff work properly.

I've bought this measurement kit: MP-1r-KIT Acoustical measurement kit

My worry is all the mention of latency and using the same input/output device.

My pc has on board sound, but I have an outboard dac (same as most here, I'd imagine) which I feed a signal via asynchronous usb.

Will this be a problem with Dirac/Acourate?

*I know that I plug the mic into the audio input on my pc's motherboard,
*get the software to make it's noises and take the measurements
*create the corrections with whichever program I choose
*and apply the convolution to JRiver

Isthat right?

Dirac has a method of matching up the timing from an input that is distinct from an output (although it sounds like it doesn't work correctly with all hardware, based on dallasjustice's experience).  According to info upthread, it sounds like even Acourate (which only supports ASIO) would work with two different devices with ASIO4All.  So it sounds like most of this software can be made to work with two different devices, it's just easier to get a predictable result if the same device is used for in and out.

The underlying technical issue is that these programs attempt to make complex and minute corrections in the time domain, by altering the timing of various frequencies (i.e. "phase").  In order to make those kinds of delicate timing corrections successfully, software needs to know exactly (to within microseconds) when the test signal "starts" in the recording. The problem is that different audio devices introduce different amounts of delay as part of their operation, and the amount of delay introduced may not be constant even for a single device over time.  At minimum, to use two different devices, software needs a way to determine what the true "start time" of the recorded signal is and potentially a way to deal with "clock drift" during the recording itself.  

The first issue is easier to fix than the second one. Any software that records an impulse response *should* be able to identify when the "start" of an impulse is (The start of an impulse is obvious enough in most cases that it could be identified manually with pretty good precision, and free measurement suites like Holm do it automatically).  If the software you're planning to use doesn't explicitly support using two devices, check and see if it a) takes impulse measurements and b) offers a way to adjust or alter the "time zero."  If both of those options are present it should be possible to work around this issue.

The second problem (clock drift during measurement) is harder to get a handle on, but given the length of most log sweeps (ten to twenty seconds), it may or may not be a huge issue (especially if you don't try to do phase correction at higher frequencies, i.e. late in the logsweep when the clock may have drifted by a significant amount).  But I've seen some pretty impressive clock drifts between devices too, so it's hard to speak in generalities.  This hasn't been a problem for me in my own measurements using two different devices, but it might be a problem for some hardware. The only way to know about the second issue, for sure, with a given set of hardware is to measure and see if there are phase anomalies in your measurement. I would expect to see a uniform, gradual rise or fall in phase over the course of the measurement which is not correlated to changes in frequency response.  I'd advise taking some measurements with your setup and see if the phase looks how you'd expect.

So while there are ways to work around it, the safest course (unless you're sure the software you plan to use supports using two different devices) is to get one of the audio interfaces mentioned upthread that can handle both input and output (unless you enjoy tinkering with measurements).  It will probably save you some heartburn down the road.

[6233638]

I'd suggest getting a mic from Herb at Cross Spectrum Labs. He calibrates Dayton mics to his professional reference. You get multiple calibration files (for using the mic at different angles) on a thumb drive and printouts of the mic's response and calibration. Good prices and I think he offers the usb mic as well.

Thanks for the recommendation, it does look like getting a mic shipped over from the US may be my best option.
 
I've spent a lot of time reading about mics now, and I'm thinking I may stretch to a SF101a microphone.
It's a lot more than I was planning on spending, but it seems like it would be worthwhile for the titanium diaphragm rather than plastic.
I have a lot of display calibration gear, and anything which used plastic filters has lost its accuracy over time (useful once profiled though) but the meters which use glass filters are still accurate and reliable.
Considering that I only plan on making this purchase once, and intend on keeping it for a long time, it seems like it would be worth paying more initially to avoid having to buy another Mic some time in the future.
 
This was probably not the best time of year to start looking into buying all this equipment, so I'm probably going to hold off on making a decision until early next year now.
If I decide not to buy the SF101a, the Audix TM1 Plus seems like a good choice - though the price for one of them here is much closer to the SF101a than in the US where it's less than half the cost - here it's more like 2/3. Either that or an Earthworks M23, which is similar in price to the TM1+.

[realysm42]

Thanks for the info guys.

Two questions:

1.So a decent mic into my PCs mobo would yield sub optimal results compared to one of these tascam devices? If thats the case, thanks for the info but sheesh, this starts to get expensive quick!

2.Reading the responses to my post might indicate a possible lack of specificity/understanding on my part; when talking about input/output devices.

Input= mic in (analogue) there to facilitate the room measurement work, regardless of software/solution.

Output= USB out (digital) the digitised version of whatever the mic picked up from the analogue input.

All of the above equates to work done on my PCs internal sound, or, one device.

When it hits my (seperate) dac via the USB, that count (to me) as another device.

Is this where the difficulties might occur, as the measurements and convolution/correction are done on the pc and sent to the dac?

I'm not sure if I'm over complicating things; in hindsight possibly yes as all the 'digital' stuff is done on this pc and then packaged and sent to the dac. I also imagine that most people using this stuff have decent, separate dacs to their PCs?

[mwillems]

Thanks for the info guys.

Two questions:

1.So a decent mic into my PCs mobo would yield sub optimal results compared to one of these tascam devices? If thats the case, thanks for the info but sheesh, this starts to get expensive quick!

2.Reading the responses to my post might indicate a possible lack of specificity/understanding on my part; when talking about input/output devices.

Input= mic in (analogue) there to facilitate the room measurement work, regardless of software/solution.

Output= USB out (digital) the digitised version of whatever the mic picked up from the analogue input.

All of the above equates to work done on my PCs internal sound, or, one device.

When it hits my (seperate) dac via the USB, that count (to me) as another device.

Is this where the difficulties might occur, as the measurements and convolution/correction are done on the pc and sent to the dac?

I'm not sure if I'm over complicating things; in hindsight possibly yes as all the 'digital' stuff is done on this pc and then packaged and sent to the dac. I also imagine that most people using this stuff have decent, separate dacs to their PCs?

The issue with two devices isn't necessarily related to outputting sound after you measure; rather it's about outputting sound to create the measurement to begin with.  When you measure using your computer, the computer has to generate and play a test signal through your speakers; you then need to record the sound of the test signal playing on the mic.  

Step 1: Software generates and begins to "play" the test signal.  In your setup, your computer sends a digital signal via USB to your DAC (bypassing your internal soundcard), and the DAC (after some amount of delay) converts the digital audio stream to analog, and sends the now analog audio signal to your amp, and your speakers then play the test signal.  There is some amount of time between when you press the "measure" button in whatever software you are using and when the test signal is actually played by your speaker (latency)

Step 2: The sound must travel from your speaker to the microphone.  This is not latency, but is actually potentially part of what you're trying to measure.

Step 3: Your microphone picks up the sound of your speakers playing the test signal and then either a) converts the analog signal to digital itself (in the case of a USB mic) or relays the analog measurement back to an analog to digital converter (ADC) in a sound card (in your case the onboard sound card's mic input).  In either case the A to D conversion involves some amount of latency, and once the analog signal is digitized it must be relayed from the audio device/driver to the software suite conducting the measurement, which also involves some amount of latency.

If you use only one audio device for both output and input, both the DAC and ADC stage are running off of the same device clock, and are using the same audio driver to communicate with the program.  This means that the latency of both step 1 and step 3 are more likely to be similar or the same, and clock drift should not be an issue.  If you use a different device for the output of the test signal than for the recording (which it sounds like you are), the latency introduced in Step 1 and Step 3 will be different (different device, different driver), and the ADC may convert the audio at a very slightly different rate than the DAC converted the digital to analog on the front end (different clock).  

If software supports using two different devices, it has some mechanism for addressing these issues.  If the software doesn't support using two different devices, correcting the different latencies is potentially possible, correcting different device clocks may be much harder (as described above), but the best way to find out if these will be a problem in your software/set up is to try and see. 

[dallasjustice]

The TASCAM us-322 can been purchased new on Amazon for $135. I bet you paid more than that for the mic and pre. You can really only try DIRAC with separate in and out devices. If I were you, I would try all 3 of them. I prefer DIRAC but you may like one of the others better. IME, the benefit is tremendous, if done right.  

Thanks for the info guys.

Two questions:

1.So a decent mic into my PCs mobo would yield sub optimal results compared to one of these tascam devices? If thats the case, thanks for the info but sheesh, this starts to get expensive quick!

2.Reading the responses to my post might indicate a possible lack of specificity/understanding on my part; when talking about input/output devices.

Input= mic in (analogue) there to facilitate the room measurement work, regardless of software/solution.

Output= USB out (digital) the digitised version of whatever the mic picked up from the analogue input.

All of the above equates to work done on my PCs internal sound, or, one device.

When it hits my (seperate) dac via the USB, that count (to me) as another device.

Is this where the difficulties might occur, as the measurements and convolution/correction are done on the pc and sent to the dac?

I'm not sure if I'm over complicating things; in hindsight possibly yes as all the 'digital' stuff is done on this pc and then packaged and sent to the dac. I also imagine that most people using this stuff have decent, separate dacs to their PCs?

[realysm42]

Thanks for the clarification Mwillems; youre right, I'll have to have a play and see what happens.

Cheers Dallas, yes I paid a fair amount more than what you've linked for me; I'll have a go with what I've got for and see how it works for me. And I'm pretty excited to see what it's going to do for me! I can only imagine the improvement it will yield (my rooms less than optimal for audio).

I'll report back once I've had a shot with it all.

[mwillems]

The TASCAM us-322 can been purchased new on Amazon for $135. I bet you paid more than that for the mic and pre. You can really only try DIRAC with separate in and out devices. If I were you, I would try all 3 of them. I prefer DIRAC but you may like one of the others better. IME, the benefit is tremendous, if done right.  

I was going to say something similar; by the time you're spending several hundreds of euros for one of these software packages, another $100 or so for an external interface doesn't seem like a huge investment.  

Also, he's not necessarily limited to DIRAC.  It took a little fiddling, but I used two different devices and managed to get good measurements using the Audiolense trial.  I made two sets of measurements: one with a microphone on the same device I was using for output (as a control), and then made one with a USB mic instead, and managed to get virtually identical measurements after some paddling around/reconfiguration.  

It's obviously not recommended (the software gave me dire warnings when I had to turn off ASIO), and I wouldn't advise someone starting from scratch to try and use two different devices for software that doesn't officially support it.  But different hardware will produce different results (which is the whole problem), and if you've already got the hardware, it's free to try the measurement part of Audiolense and see if you get flaky results with two devices.

[realysm42]

@Mwillems, I'm sorry I'm really new to this side of audio.

You're agreeing with Dallas by saying the tascam is superior to the mic input on a mobo for taking measurements?

[mwillems]

@Mwillems, I'm sorry I'm really new to this side of audio.

You're agreeing with Dallas by saying the tascam is superior to the mic input on a mobo for taking measurements?

Yes.  Almost any sound device will be superior to Mobo sound, and having one device for both input and output during measurement will make your life easier.

The extra device will cost you something, but most of the software packages being discussed here are themselves pretty expensive (between $250 and $650). 

[realysm42]

I understand with the kit I have it works like this:

Mic > mic pre > mobo (via 3.5mm jack) = recording

I'm not sure I understand how the tascam would fit into the equation, I can only guess it would look like this:

Mic > tascam pre > mobo (via USB) = recording

And the benefit of the tascam is not using the headphone input?

Thanks for your explanation!

[mwillems]

I understand with the kit I have it works like this:

Mic > mic pre > mobo (via 3.5mm jack) = recording

I'm not sure I understand how the tascam would fit into the equation, I can only guess it would look like this:

Mic > tascam pre > mobo (via USB) = recording

And the benefit of the tascam is not using the headphone input?

Thanks for your explanation!

That's about right, but I wanted to clarify one thing.  The key difference between those two signal chains is that the TASCAM would be handling the Analog to Digital conversion (ADC) and sending a digital signal to the computer via USB, instead of your current setup which sends an analog signal to the analog to digital converter built into the motherboard (the MoBo jack).  Think about the quality of the DAC on the MoBo soundcard as compared to the outboard DAC you have.  The ADC in the MoBo soundcard is the same way, an outboard ADC is likely to be much better.

The other advantage is that, if you pick an external device with both a DAC and an ADC in it (like the TASCAM), you can use it to handle both sides of the measurement process (and switch back to your existing DAC when you're done measuring).

[dallasjustice]

The tascam also can go out as well.  So, you plug the tascam into the usb.  The tascam has a mixer which also has a control panel that allows you to configure it to send 2 channels out the back of the tascam into your amp while also using the mic input to go into the tascam.  This way the output and the input are controlled by the clock inside the tascam unit.  Thus, the measurement will be synchronous.  

One word of caution:  I have done this configuration.  If you go directly to your amp out of the tascam, make sure you know how the volume control works in the tascam mixer.  I like to use a couple of 12db resistors between the tascam and my amps.  That way, I can confirm the volume is working correctly before I send a test signal through.  I have blown a tweeter in alll of this.  :-)  I had to send my speaker back to the manufacturer to have the tweeter replaced and remeasured.  Kinda takes the fun out of it for a short time.  

I understand with the kit I have it works like this:

Mic > mic pre > mobo (via 3.5mm jack) = recording

I'm not sure I understand how the tascam would fit into the equation, I can only guess it would look like this:

Mic > tascam pre > mobo (via USB) = recording

And the benefit of the tascam is not using the headphone input?

Thanks for your explanation!

[mwillems]

One word of caution:  I have done this configuration.  If you go directly to your amp out of the tascam, make sure you know how the volume control works in the tascam mixer.  I like to use a couple of 12db resistors between the tascam and my amps.  That way, I can confirm the volume is working correctly before I send a test signal through.  I have blown a tweeter in alll of this.  :-)  I had to send my speaker back to the manufacturer to have the tweeter replaced and remeasured.  Kinda takes the fun out of it for a short time.  

Yikes, that's no fun.  Tweeters are delicate creatures. I've been lucky so far, but there were one or two episodes where I was sure I'd fried them (but thankfully hadn't). 

[realysm42]

Thank you both for your input and information, gentlemen.

When you explain about the tascam as a superior ADC, it makes perfect sense to me; basically, the less the mobo has to do with the actual signal processing, the better!

Hmm, thats pretty scary, I've got a beast of an amp (500w RMS into 8ohms, 900 into 4ohms). And my speakers aren't cheap or easy to move (40kg+ each) so blowing a tweeter isn't an option for me. You used physical attenuation?

Can't you just turn it right down to start with and carefully increase volume until you're at a safe level?

[dallasjustice]

Can't you just turn it right down to start with and carefully increase volume until you're at a safe level?

You would think so.   

[mojave]

The other advantage is that, if you pick an external device with both a DAC and an ADC in it (like the TASCAM), you can use it to handle both sides of the measurement process (and switch back to your existing DAC when you're done measuring).

I use the Tascam for my mic input (ADC only), but my regular DAC for output. However, in Audiolense I am only picking one input and output device. Do you know how I do it? 

Can't you just turn it right down to start with and carefully increase volume until you're at a safe level?

That is what I do. I haven't had any issues with the Tascam. Also, many measurement programs let you attenuate digitally, too.

[mwillems]

I use the Tascam for my mic input (ADC only), but my regular DAC for output. However, in Audiolense I am only picking one input and output device. Do you know how I do it? 

I bet I've got an idea 

Out of curiosity, any specific reason you don't just use the mic inputs on the UR824 directly?  Is it just convenience (i.e. cable lengths, UR824 in an out of the way rack, etc.) or do you just like the ADC in the TASCAM more?

[mojave]

I bet I've got an idea 

Out of curiosity, any specific reason you don't just use the mic inputs on the UR824 directly?  Is it just convenience (i.e. cable lengths, UR824 in an out of the way rack, etc.) or do you just like the ADC in the TASCAM more?

Last April I purchased a two channel AES DAC and an 8 channel AES DAC from a local designer (Ross Martin Audio). I also bought the Lynx AES16e. I wanted to try something else out and have more channels available in case I decided to go active with my mains. With this system I needed a preamp for my mic so I bought the Tascam. The Tascam has AES output so the Lynx AES16e is the ASIO input/output device in Audiolense. I wanted to try everything out and evaluate the measurement ability and playback quality before making any changes. About a month ago I finally changed all my cables from TRS/XLR to XLR/XLR and swapped gear.

So, I am no longer using the UR824 in my system. The UR824 did have a higher SNR in the measurement according to Audiolense than the Tascam and is much easier to use for measurements. Probably the only thing "better" about the AES DACs is the sound quality and lower latency (for loopback). I really think the Steinberg is the best overall audio device I've ever used.

I just received an e-mail that next week I will be getting a demo Solid State Logic Alpha-Link MX and MadiXtreme card. This has 16 channels of output for $1699 and you can later expand to 32 channels.

[dean70]

Anyone used the Motu UltraLite-mk3 Hybrid? It has Mic ins and 10 x balanced analog out channels & is reasonably priced. Cannot find any technical specs on the unit.