V5.20.14 early access build

[John Mulcahy]

John, I am trying the new 64k short sweeps and impulse responses are shifting from measurement to measurement by around 7mm. None of the 128k sweeps deviate from each other. Since 64k sweeps should be handling USB mic timing problems even better, I suspect something is wrong with the new 64k sweeps. It might be coincidence but as you'll see I took 11 consecutive 128k measurements with not one deviation whereas every other 64k sweep moved.

The capture time could be slightly too short sometimes when using the acoustic reference, I have fixed it for the next build.

 

[John Mulcahy]

Would there be any way to make it possible to select multiple measurements using CTRL/SHIFT, or even by checking a box, and using a command like "Export Selected"?

There are export options for the selected measurements in the All SPL graph controls.

 

[dcibel]

I wonder if there's the possibility of fixing the fact that the cursor location drifts as you scroll, so the cursor doesn't stay put if you adjust the view at all.
Example, cursor located top right of window, at 2kHz and -71.26dB. Scroll the window so same location is shown in bottom right, without touching the cursor, it has drifted to 1.969kHz and -70.8dB

 

[John Mulcahy]

I wonder if there's the possibility of fixing the fact that the cursor location drifts as you scroll

Fixed for the next build.

 

[John Mulcahy]

Builds updated today with these changes:

• Added: Double-click on graph brings up the Limits dialog
• Changed: Altered room sim to give more consistent response levels as room volume changes
• Fixed: Capture time needed to be longer when using 64k sweep with acoustic timing reference
• Fixed: Cursor value could change as graph was scrolled
• Fixed: initialprefs.txt in the REW log files folder was not used
• Changed: If rewprefs.txt is found in the REW log files folder it will be used to store preferences, unless the -prefs argument or -Drew.preferences.file roomeqwizard.vmoptions entry have been specified. That means switching to file-based preferences can be done quickly by saving preferences to rewprefs.txt in the log files folder and restarting REW. On Windows systems startup is faster with file-based preferences.

 

[Dilon]

Is it possible to to have additional support for EQ's from manufacturer as:

IOTAVX AVX17
t.racks dsp 4x4 mini
Behringer Inuke dsp

 

[John Mulcahy]

There is a configurable equaliser option to help accommodate additional devices.

 

[sam_adams]

@John Mulcahy, I just want to bring to your attention some issues that have cropped up with users installing the 32-bit version on a 64-bit Windows installation. In the big REW discussion over on AVS, this post indicates that the user's issues were due to a 32-bit install on a 64-bit OS. Also, you just recently dealt with this user's issue regarding a 32-bit install on a 64-bit OS. If it is possible; could the installer be built in such a way as to check what version of Windows the installation is being run on and display a warning to notify the user of such? It may just spare some users from frustration and reduce the number of questions that result from an incorrect installation.

 

[John Mulcahy]

I'll add an architecture check in the next build. It can't be done in the installer as the processor family and CPU need to be checked along with the architecture which I can't easily do within the installer. There are some old 64-bit processors which exhibit sporadic crashes with a 64-bit Java runtime, REW recommends the 32-bit build if it detects it is running on one of those.

 

[serko70]

John,

I have been getting clock delays with "from file" sweeps of 64k as below:



Impulse peaks are also some 27cm apart with the "from REW" sweeps

I guess your fix is not yet implemented in the generator.

 

[John Mulcahy]

The change was to the capture side, the generator isn't affected. If you attach the files I can have a look.

 

[serko70]

The change was to the capture side, the generator isn't affected. If you attach the files I can have a look.

 

[serko70]

John, I found the problem. The speaker windows sample rate setting was left at 192kHz hence the delays. Sorry , nothing to fix!

EDIT:

I still get it with Java drivers (EXCLS included), doesn't happen with ASIO, here's another mdat:

 

[achugh]

Has anyone else experienced that the SPL Meter button after a fresh install will not play the sound from the LFE channel from the last 10-15 beta build? On the Choose signal source window you will have to change the dropdown value from the default "Use REW subwoofer cal signal" to "Use REW speaker cal signal" and then back to "Use REW subwoofer cal signal" couple of time for the sound to finally start coming out of the LFE channel?

Once it starts working, it works until you do not change your sources, or your sources are automatically changed because REW detects a change. Then you are faced with this problem again. You fiddle through this dropdown till it starts outputting sound. Every once in a while I also have to go into preferences to set the output to use LFE channel shutdown REW and restart and then come back to this window and fiddle some more for REW to then start sending the LFE sound.

 

[John Mulcahy]

I have fixed that for the next build. As a workaround in the meantime open the generator and start the cal signal before pressing the Calibrate button on the SPL meter.

 

[JoinUs]

When using the Alignment tool, I would like to be able to adjust the slider smoothly with the mouse scroll wheel, as is currently available in beta version EQ. Or if that's not possible, I'd like to be able to adjust it with the keyboard's direction key.
Currently, you must type every time or adjust the slider yourself.

 

[John Mulcahy]

You can make fine adjustments using the left/right arrow keys. I'll add mouse wheel adjustment in the next build, that has coarse and fine adjustments where fine is accessed by holding the Alt key while scrolling.

 

[sm52]

John Mulcahy​

I have made many measurements in the car using the CD-recorded signals from the REW generator. Automatic clock correction always gives slightly different numbers. In my case, from -83 ppm to -87 ppm. Can a CD player's clock float that much? I assumed this was unlikely and tried to figure out some 'correct value' of the clock correction from several measurements using the clock correction in the impuls window. I got 86.6 ppm. Can REW make inaccurate clock corrections for some reason?

 

[John Mulcahy]

-83 to -87 is a pretty narrow range, within the variation that might be seen from a device. Clock correction depends on the location of the timing signals, accurately locating them may be affected by strong nearby reflections which are the norm for a car.

 

[sm52]

How do you advise to do -
- rely on automatic REW clock correction or
- manually figure out the 'exact value' of the correction and apply to all measurements or
- calculate the average correction value and apply to all measurements?

 

[John Mulcahy]

Clocks vary over time, there is no exact value. Either use the REW figure or turn off the correction if it seems unreliable.

 

[sm52]

Thanks for the answer. I took measurements with the correction turned off. The distances between the impulses are exactly the same as with the included correction. But the phase is different. If you need to match two drivers in phase and use an alignment tool, clock correction is required.

 

[John Mulcahy]

Builds updated today with these changes:

• Added: Airy and Morlet continuous wavelet transform modes for the spectrogram graph
• Added: Export buttons for the decay, waterfall and spectrogram graphs
• Added: Export IR as WAV option to export a specified number of samples
• Added: IR windows dialog display of number of samples the windows span
• Added: Mouse wheel can be used to adjust sliders for the alignment tool and t=0 offset
• Added: Recommend the 64-bit build on 64-bit architectures
• Fixed: Generator might not start when attempting SPL calibration

The Airy CWT and Morlet CWT modes are Java implementations of a continuous wavelet transform using the algorithm described in Arts, L., & van den Broek, E. (2022). The fast continuous wavelet transformation (fCWT) for real-time, high-quality, noise-resistant time–frequency analysis. Nat Comput Sci, 2(1), 47–58. https://doi.org/10.1038/s43588-021-00183-z

The Morlet wavelet is commonly used for CWT, though it has some deficiencies at high time resolutions, producing artefacts which are a consequence of it being only approximately analytic. The Airy wavelet is from the γ = 3 family of the generalised Morse wavelets. It is exactly analytic and has better performance than the Morlet wavelet at octave fractions from 1/1 to 1/5. At high frequency resolutions, octave fractions from 1/12 to 1/24, the Airy wavelet produces artefacts and the Morlet wavelet is a better choice.

REW's Wavelet mode implements the transform using complex smoothing and the time shift properties of the FFT, with a kernel that is less gaussian (tending more towards triangular) than Morlet or Airy. That gives sharper definition to higher level features of the plot at the expense of more low level artefacts.

 

[serko70]

Thank you John. I ran into a little problem. Dirac pure pulse created from empty filters comes at 96kHz and cannot be used in TA with other measurement.

 

[John Mulcahy]

Thank you John. I ran into a little problem. Dirac pure pulse created from empty filters comes at 96kHz and cannot be used in TA with other measurement.

Generate it from a measurement at the sample rate you want to use.