REW Beta Release REW API beta releases

[Breeman]

Yes. The filters aim to bring the response to the target, so their response is drawn centred about the target level. The target normally follows the target level, unless you are using a house curve which is not relative to zero.

I am trying to get the target (house) curve, the blue dashed line shown below and the Filters response graph (shown in the second picture below), to be visible on the same screen so I can click and drag to adjust the individual filters, to EQ the subwoofer's response towards the target curve below 100 Hz.
Currently, if I set the Target Level (dB SPL) to 0 dB, the adjustable Filter responce graph gets pushed down by 75dB such that it moves off screen as shown in the third picture below. I want to be able to have two graphs, the target and the filters curves, visible in the same screen. Is this doable?

I have attached the target curve for your reference.

 

[serko70]

I am trying to get the target (house) curve, the blue dashed line shown below and the Filters response graph (shown in the second picture below), to be visible on the same screen so I can click and drag to adjust the individual filters, to EQ the subwoofer's response towards the target curve below 100 Hz.
Currently, if I set the Target Level (dB SPL) to 0 dB, the adjustable Filter responce graph gets pushed down by 75dB such that it moves off screen as shown in the third picture below. I want to be able to have two graphs, the target and the filters curves, visible in the same screen. Is this doable?

I have attached the target curve for your reference.

View attachment 72762


View attachment 72763

View attachment 72764

Click the "Limits" toolbar at the top right corner and select "Fit to Data". You can also minimize Waterfalls graph to have a larger EQ window.

 

[John Mulcahy]

I am trying to get the target (house) curve, the blue dashed line shown below and the Filters response graph (shown in the second picture below), to be visible on the same screen so I can click and drag to adjust the individual filters, to EQ the subwoofer's response towards the target curve below 100 Hz.

Your house curve file has a 71 dB offset. Make one that is zero dB at some reference frequency, e.g. 1 kHz.

 

[John Mulcahy]

Knowing how much the graphs have been shifted is useful outside of REW. Can I see it somewhere?

Cumulative IR shift on the Info panel.

 

[sm52]

Cumulative IR shift on the Info panel.

I found. Thank you.
I checked the measurement of 1.5 µF capacitors. The component model is created. The names of the calibration steps are in place. Thank you.

 

[Breeman]

Your house curve file has a 71 dB offset. Make one that is zero dB at some reference frequency, e.g. 1 kHz.

Sorted, thanks John.

 

[Tikkidy]

REW 5.4 Beta 48:

I wonder what's causing the apparent rise in distortion for 64K samples x8 repetitions?

 

[John Mulcahy]

I wonder what's causing the apparent rise in distortion for 64K samples x8 repetitions?

From the help:

Impulse responses measured using logarithmic sweeps separate distortion from the linear part of the system response. The distortion components appear at negative times, behind the main impulse. Analysing the frequency content of these components allows plots of distortion harmonics to be generated. The longer the sweep, the better the distortion components are separated from each other. Sweeps of 256k or shorter should start at 0 Hz to prevent an artificial rise in distortion at the lowest frequencies due to the reduced low frequency bandwidth of the resulting impulse response. When measuring a system with high distortion levels use a long sweep setting (e.g. 1M or higher), at shorter sweep lengths the harmonics may affect each other giving misleading results.

TLDR; If you are interested in distortion use long sweeps.

 

[Tikkidy]

RTFM.

Thanks John!

I’ve noticed that a 1 or 2M sweep at 44.1 or 88.2K sample rate gives good signal to noise ratio. And a very long sweep + multiple repetitions eg. 4Mx8 can give an improved SNR of some 20dB, which helps resolve harmonics. The downside of this, is that when testing acoustic devices at higher voltages, it runs the risk of overheat voice coils and causing permanent damage to drivers (4M x8 is a 6+ minute test!)

A 256K x8 is close to 2M in terms of signal to noise, but more helpful when measurements are taken in a non-anechoic environment. I find it preferable because it is able to reject implosive environmental sounds that may otherwise disturb the measurement.

So in practice, I find using 256K x8 gives the best balance of ability to resolve harmonics.

For tweeters, it’s important to the keep the sweeps as short as possible eg. 3 seconds or less, in order to avoid overheating

 

[Tikkidy]

Also, the latest stable release of REW, I could shift the fundamental to denote "measured at 0.5m, plotted for 1m"
Where has Actions now moved to?

 

[John Mulcahy]

Right click on the graph and on the measurements for additional actions.

 

[John Mulcahy]

Builds updated today (beta 49) with these changes:

• Added: The RTA distortion settings have a "Cross correlation averaging" option to reduce input noise when measuring a signal that is connected to both the measurement and reference inputs
• Added: Waterfall and spectrogram have an "Apply settings to all" button to replicate settings to other measurements
• Changed: Draw filter responses relative to the total target response level at 1 kHz rather than relative to the target level figure
• Changed: When the "Enable mousewheel zoom" View preference is not selected the mouse wheel may still be used to zoom by scrolling on the axes
• Fixed: Graph smoothing shortcuts on macOS did not work properly
• Fixed: Dither bit depth box was missing for tone generator

 

[John Mulcahy]

Builds updated today (beta 50) with these changes:

• Added: Stereo WAV files dropped on the RTA can be processed using cross correlation averaging
• Added: Cross correlation averaging can plot the magnitude of the complex result as an alternative to the real part of the result
• Added: Speech noise signal per IEC 60268-16:2020
• Fixed: RTA option to stop at a number of averages should not be active for file data
• Fixed: Saving noise signals to file from the generator with two channels saved correlated noise even if uncorrelated was selected
• Fixed: Clicking Calibrate on the SPL meter with a calibrated UMIK on macOS would request a cal file when it should say cal is not required

 

[KSTR]

Fixed: RTA option to stop at a number of averages should not be active for file data

Oh, I thought that was a very nice option and I've used it quite some times.
One can always stop the averaging manually during realtime processing so it's sort of redundant for this.
OTOH, for files you have no other way than to press "cancel" and because the progress indicators counts frames, not averages, it is pretty hard to stop at an even approximate only number of averages, especially for small FFT sizes.

 

[John Mulcahy]

I can look at making the stop feature work for files, but it wasn't working correctly as it was.

 

[carl second]

When adjusting the SPL dB level offset of an imported rePhase generated filter down to zero, one needs to apply "Add to data" twice otherwise the SPL offset adjustment doesn't seem to permanently register.

I've also noticed that in the Overlays predicted phase view window "phase wrap lines" -- if unchecked in the view settings -- still continue to appear.

 

[John Mulcahy]

I have fixed those for the next build.

 

[sm52]

John Mulcahy, hello. A few questions. I have several measurements. I processed one of them in the equalizer window and saved the processed measurement. In particular, I reduced one peak. After this, the measurement with the equalizer stopped displaying distortion. There are no graphs for it in the distortion window. This is right?

In the Overlays window, distortions, N and THD+N are not displayed for sweep measurements. This is right?

If there are multiple measurements that have the same noise elements above the underlying noise level. Some kind of interference from the power supply. Can REW, through cross-correlation or another method, detect this and remove it (reduce it to the main noise level)?

 

[John Mulcahy]

I processed one of them in the equalizer window and saved the processed measurement. In particular, I reduced one peak. After this, the measurement with the equalizer stopped displaying distortion. There are no graphs for it in the distortion window. This is right?

Yes, that's right. You could use "Measure with these filters" to see their effect.

In the Overlays window, distortions, N and THD+N are not displayed for sweep measurements. This is right?

Yes, that's right. Sweep measurements have no equivalent to the single-frequency N and THD+N parameters.

If there are multiple measurements that have the same noise elements above the underlying noise level. Some kind of interference from the power supply. Can REW, through cross-correlation or another method, detect this and remove it (reduce it to the main noise level)?

No, but if you have a means of generating an inverted copy of the input signal and applying it to a second input channel the cross-correlation processing with the Invert correlation reference option may allow common mode noise on the inputs to be cancelled. There is a thread on cross-correlation processing here.

 

[sm52]

Thanks for the answers. Your third answer concerns the measurement of electrical signals, but acoustic measurements can also be carried out if the conditions are met?

 

[John Mulcahy]

That would probably be pointless. For acoustic measurements an analog mic usually carries a differential signal that would automatically reject common mode noise. Remaining unwanted content is likely to be acoustic and so part of the signal. For sweep measurements using the same interface for input and output multiple sweeps would help to reduce interference.

 

[sm52]

Thank you. It's clear.

 

[sm52]

if you have a means of generating an inverted copy of the input signal and applying it to a second input channel the cross-correlation processing with the Invert correlation reference

I'm trying to figure out what the electrical connection should look like to do this. There is an input for a signal, there is an input for a loop from output to input. These are two inputs. Where should the inverted input signal be sent? Need a third input?

 

[Breeman]

Hi John,

I am trying to align the impulse responses of the same subwoofer, one measurement before EQ and another after EQ. I want to use the impulse response of the EQ'd measurement as a reference but cannot easily do it without having to click Apply & prev or Apply & next buttons, using the Offset t=0 function under Overlays. I think there is a use case to be able to select or change (via a dropdown or similar) a reference measurement from within the offset adjustment popup against which to align the impulses of other measurements.

I also noticed that the phase response shown is that of the measurement being time adjusted. I think it would a good idea to be able toggle the phase response of the reference measurement on and off, in case one is interested to see phase alignment.

A option to invert the impulse response of the selected measurement from within the popup would be appreciated too.

 

[Breeman]

Hi John,

I am trying to align the impulse responses of the same subwoofer, one measurement before EQ and another after EQ. I want to use the impulse response of the EQ'd measurement as a reference but cannot easily do it without having to click Apply & prev or Apply & next buttons, using the Offset t=0 function under Overlays. I think there is a use case to be able to select or change (via a dropdown or similar) a reference measurement from within the offset adjustment popup against which to align the impulses of other measurements.

I also noticed that the phase response shown is that of the measurement being time adjusted. I think it would a good idea to be able toggle the phase response of the reference measurement on and off, in case one is interested to see phase alignment.

A option to invert the impulse response of the selected measurement from within the popup would be appreciated too.
View attachment 73000

Hmm, I just realised that the same thing can be accomplished using the Alignment tool. However there are no Apply & next or prev options on the Alignment tool.