Member
Joined 2003
I should clarify, running multiple sweeps and averaging in ARTA does appear to be an option, it does not appear to be an option in REW. I'll find some time soon to run a few sweep tests to see how the noise level is affected in distortion analysis. Maybe tomorrow.
Yes. Running multiple sweeps is the equivalent of averaging.
So the noise goes down with a 10 log (x).
In other words you need a lot averages to improve anything significant after a while.
The distortion mainly depends on the ADC and preamp (as well as the DAC obviously).
Computing distortion takes close to nothing these days.
So the noise goes down with a 10 log (x).
In other words you need a lot averages to improve anything significant after a while.
The distortion mainly depends on the ADC and preamp (as well as the DAC obviously).
Computing distortion takes close to nothing these days.
Member
Joined 2003
I compared steps to ARTA sine sweep directly today, and couldn't conclude that one method was better over the other. They both achieved roughly the same result, with the exception of course that the sine sweep took only a couple seconds, verses a couple minutes for STEPS.
The only different I saw, was the driver I tested has a distortion peak centred around 900Hz, which appeared a bit of a wider peak with the sine sweep than with steps. Otherwise the harmonic distribution and amplitude was the same between both.
The only different I saw, was the driver I tested has a distortion peak centred around 900Hz, which appeared a bit of a wider peak with the sine sweep than with steps. Otherwise the harmonic distribution and amplitude was the same between both.
It is definitely possible in REW.
I did it last week actually trying out some EQ methods.
The "quick" method accuracy depends on how polluted the impulse is by artifacts like room reflections and noise. If you look in the ARTA Docs you will find this passage:
STEPS is better in this regard but it does take a relatively long time.
Member
Joined 2003
Does anyone know if it's possible to take "time delayed" measurements with ARTA ?
What I mean by that is normally measurements are taken in real time with a sound card where it outputs the test signal (sweep, noise etc) and simultaneously records the result and analyses it.
However I have devices whose frequency response I'd like to measure (not worried about absolute phase) which I can't feed a live analogue signal into.
For example I'd like to measure the frequency response of the headphone output of my TV, in different sound modes.
I'd also like to measure the (acoustic) frequency response of the stereo in my car, mainly in the bass region.
Neither of these have analogue inputs which I could connect the USB sound interface I use with ARTA to.
However both have mechanisms to play back bit perfect recorded digital files on them such as WAV etc through various means.
So what I'm looking for is to set up my measurement options as normal (noise, sweep, FFT size etc) then instead of doing the measurement real time, hit a button and have ARTA generate a lossless file of some sort (WAV, FLAC etc) which I can then copy onto a thumb drive or use in some way to play back on a device that doesn't support analogue input.
I could then play the pre-recorded test signal on the device with ARTA recording the result for it to analyse. Obviously absolute phase cannot be determined using a technique like this but a minimum phase response can be calculated.
Yes I could probably use the continuous RTA mode which just puts out repetitive noise which could be captured and played back into it in RTA mode, however that is not quite the same as generating an impulse response in terms of what you can do with the result. (For example noise/distortion measurements, other stuff like ETC response and so on...) The RTA mode is fairly limited.
Am I missing an obvious way to do measurements where the test signal is exported as a file and played back to ARTA at a later time for analysis ?
There are many test scenarios and devices where this delayed, two part measurement process would be useful.
What I mean by that is normally measurements are taken in real time with a sound card where it outputs the test signal (sweep, noise etc) and simultaneously records the result and analyses it.
However I have devices whose frequency response I'd like to measure (not worried about absolute phase) which I can't feed a live analogue signal into.
For example I'd like to measure the frequency response of the headphone output of my TV, in different sound modes.
I'd also like to measure the (acoustic) frequency response of the stereo in my car, mainly in the bass region.
Neither of these have analogue inputs which I could connect the USB sound interface I use with ARTA to.
However both have mechanisms to play back bit perfect recorded digital files on them such as WAV etc through various means.
So what I'm looking for is to set up my measurement options as normal (noise, sweep, FFT size etc) then instead of doing the measurement real time, hit a button and have ARTA generate a lossless file of some sort (WAV, FLAC etc) which I can then copy onto a thumb drive or use in some way to play back on a device that doesn't support analogue input.
I could then play the pre-recorded test signal on the device with ARTA recording the result for it to analyse. Obviously absolute phase cannot be determined using a technique like this but a minimum phase response can be calculated.
Yes I could probably use the continuous RTA mode which just puts out repetitive noise which could be captured and played back into it in RTA mode, however that is not quite the same as generating an impulse response in terms of what you can do with the result. (For example noise/distortion measurements, other stuff like ETC response and so on...) The RTA mode is fairly limited.
Am I missing an obvious way to do measurements where the test signal is exported as a file and played back to ARTA at a later time for analysis ?
There are many test scenarios and devices where this delayed, two part measurement process would be useful.
Last edited:
There is external excitation mode in ARTA ?
Or is that only a trigger function thing?
But in general I think you can use whatever.
Because you can also always import external impulse files as well.
Even in .wav format.
So basically you can let your device play some kind of burst or sweep.
Record it, maybe clean it up with something like Audacity and you're good to go.
That would also work in VituixCAD btw.
Which also has a few more window gate options.
Or is that only a trigger function thing?
But in general I think you can use whatever.
Because you can also always import external impulse files as well.
Even in .wav format.
So basically you can let your device play some kind of burst or sweep.
Record it, maybe clean it up with something like Audacity and you're good to go.
That would also work in VituixCAD btw.
Which also has a few more window gate options.
Member
Joined 2003
Open the help file in ARTA and read section 2.12 "Saving Generator Signals in a .wav file"
Then you can use ARTA as you wish, select "external" as the generator source.
Thanks, but that's not what I'm looking for.
That is still just the RTA (real time analyser) mode but not in real time, eg saving the excitation signal to a file then playing it back later to be captured.
I'm looking for some way to use the "Impulse Response / Signal Time Record" mode via a saved WAV file, as the RTA mode is quite limited in functionality compared to the Impulse response mode. (For example you can't use swept sine for excitation like you can in Impulse Response mode, nor are many of the different measurement types aside from frequency response available)
Last edited:
Member
Joined 2003
Simple answer to your inquiry on impulse response measurement from external source - ARTA won't do it.
Impulse response is time sensitive function, which is why ARTA is set up to measure these signals in real time duplex mode. For the use cases you mention, TV headphone output and car stereo, impulse response is simply not needed, frequency response can be determined using the noise signals and RTA, which are not time sensitive measurements.
Impulse response is time sensitive function, which is why ARTA is set up to measure these signals in real time duplex mode. For the use cases you mention, TV headphone output and car stereo, impulse response is simply not needed, frequency response can be determined using the noise signals and RTA, which are not time sensitive measurements.
This is quite misleading - the only thing that would be lost measuring an impulse response in a record/playback scenario where there is a long uncertain time delay instead of real time is absolute time of flight information.
This means absolute phase and absolute group delay can't be measured as you have no time reference, however everything else which doesn't depend on knowing the time of flight - which is the majority of the measurement types you can make based on the impulse measurement would be fine.
It isn't implemented at the moment but there's no reason it couldn't be.
You keep going on about frequency response as if that is the only measurement that exists but there are so many other measurements you can derive in the impulse response mode that you can't in the RTA mode in ARTA.
I gave the example of distortion measurements already. (Analysis->Frequency response and distortions)
There is also step response, ETC curve, energy decay, CSD and Burst delay to name a few. All these measurements can be taken without absolute time references and are perfectly useful without absolute time references. (What relevance does absolute time reference of the measurement in a CSD have for example ?)
With certain signal excitation types its not even difficult for ARTA to scan for and locate the excitation signal in a several second capture period - for example the sine sweep mode already has "centre peak of impulse response" as an option, which will automatically adjust the time reference to the impulse peak, nulling out variations in time of flight giving a more consistent measurement to measurement alignment with the start of the test excitation signal.
Being able to use most of the features provided by the Impulse measurement mode via delayed recording and playback is absolutely possible if Ivo decided to add that feature, I can't see anything that would prevent it working, allowing for the fact that you won't have an absolute time/phase reference for any of the measurements.
As you say, it doesn't appear to have this feature at the moment though - which is why I asked about it in the first place.
Other than time taken for testing, is there any other problem with the stepped-sine method? If I can always afford the time (I'm a hobbyist, I always can) should I always use STEPS instead of the swept sine?
Any recommendations about the parameters for STEPS to get more accurate results? Like the beep duration, the delay between beeps, etc?
I also really wish I could apply smoothing to the extremely "hairy" distortion plots which STEPS generates.
But my question was whether it could be done with ARTA and if so how, not what other software could I use besides the software I paid for, know and love, and have used for many years for audio measurements. After all this is the ARTA thread I posted in...
As for "MidFi TV outputs" that was just an example. I can think of numerous devices whose DAC and analogue outputs could be measured which do not have analogue inputs to allow real time measurement. So there is a fairly wide use case.
Perhaps it also doesn't occur to you that measuring devices of "unknown" and potentially middling quality level might be why the measurements are being taken in the first place, to establish what their true level of quality is ?
There's no fun in measuring something that measures almost perfect...
Sorry for the OT... I'm obviously too used to look for the easiest method (for me, obviously) to perform a task and not enough patience to go to leghths with a tool requiring workarounds, just because I have it used before.
BTW, even the tool I mentioned can measure way lower than anyone can hear and below any threshold where a device could have a negative impact.
BTW, even the tool I mentioned can measure way lower than anyone can hear and below any threshold where a device could have a negative impact.