V5.20.14 early access build

[John Mulcahy]

In the latest version, "Measurement repeats:3" setting is being ignored in both single and sequential measurements. Measuring only once.

Yes, repeats is for the repeated measurements mode. Single is, as the name says, a single measurement. Sequential measures a selected sequence of channels.

Did the first acoustic reference chirp tone change and get shorter?

No.

Do we still need to untick "Decimate IR" in preferences to keep LFE measurement sampling rate at 48kHz when we use LFE end frequency during measurement?

Yes.

 

[sam_adams]

@John Mulcahy, the below might be visual anamolies (ea52 on macos 13.3.1 and below):

In the EQ Filters section, when the EQ Filters pane is opened the filters display like below:

unless/until either the 'None' button or the 'Auto EQ' text is clicked on and then they display correctly:

In the Alignment tool, when clicking on the left side (lower frequency) in the frequency plot in the top section, the span of the Impulse plot increases:

When clicking further to the right in the frequency plot section, the Impulse plot span becomes shorter:

Bug, feature, or by design?

 

[John Mulcahy]

Both by design. Per the first post in this thread, the Show all/Hide all buttons toggle the filter control views. The state for an individual filter can be changed by left clicking it to show the controls, right click or double click hides them. Mouse wheel control is available in either view. if you don't like the compact view use the Show all button, it will be remembered for the next startup.

The alignment tool impulse span depends on frequency, maintaining a consistent number of cycles of the alignment frequency.

 

[sam_adams]

Both by design. Per the first post in this thread, the Show all/Hide all buttons toggle the filter control views. The state for an individual filter can be changed by left clicking it to show the controls, right click or double click hides them. Mouse wheel control is available in either view. if you don't like the compact view use the Show all button, it will be remembered for the next startup.

The alignment tool impulse span depends on frequency, maintaining a consistent number of cycles of the alignment frequency.

Excellent! Thank you for clarifying.

 

[serko70]

REW ea52 threw NullPointerException when I unplugged USB mic and DAC:


REW V5.20.14 running Azul Systems, Inc. JRE 1.8.0_362 64-bit on Windows 10 Locale: language en, keyboard GB, windows-1252 at 96 DPI


Message:
java.lang.NullPointerException

Spoiler: Stack trace

Stack Trace:
roomeqwizard.D$7.actionPerformed(y:2766)
javax.swing.Timer.fireActionPerformed(Timer.java:313)
javax.swing.Timer$DoPostEvent.run(Timer.java:245)
java.awt.event.InvocationEvent.dispatch(InvocationEvent.java:311)
java.awt.EventQueue.dispatchEventImpl(EventQueue.java:758)
java.awt.EventQueue.access$500(EventQueue.java:97)
java.awt.EventQueue$3.run(EventQueue.java:709)
java.awt.EventQueue$3.run(EventQueue.java:703)
java.security.AccessController.doPrivileged(Native Method)
java.security.ProtectionDomain$JavaSecurityAccessImpl.doIntersectionPrivilege(ProtectionDomain.java:74)
java.awt.EventQueue.dispatchEvent(EventQueue.java:728)
java.awt.EventDispatchThread.pumpOneEventForFilters(EventDispatchThread.java:205)
java.awt.EventDispatchThread.pumpEventsForFilter(EventDispatchThread.java:116)
java.awt.EventDispatchThread.pumpEventsForHierarchy(EventDispatchThread.java:105)
java.awt.EventDispatchThread.pumpEvents(EventDispatchThread.java:101)
java.awt.EventDispatchThread.pumpEvents(EventDispatchThread.java:93)
java.awt.EventDispatchThread.run(EventDispatchThread.java:82)

 

[John Mulcahy]

You managed to unplug the mic just after a check that there was one and just before the following line that uses its input volume setting. I'll add something to deal with that.

 

[AudiocRaver]

John,

On the Generator page, ver. 5_20_14ea52, with Octave periodic pink noise selected, when scrolling through the selectable center frequencies from lower to higher, the 63 Hz center frequency covers the band from 35 Hz to 89 Hz, and when scrolling from higher to lower, it covers the correct band from 45 Hz to 89 Hz. Probably just a code typo somewhere.

Red: Pink Noise, 63 Hz octave, scrolling up through center freq selector, 35 to 89 Hz
Green: Pink Noise, 63 Hz octave, scrolling down through center freq selector, 45 to 89 Hz

As always, thank you for all your tireless work on this amazing tool!

 

[John Mulcahy]

Thanks for that. The problem is premature intrusion of a minimum bandwidth check as the low and high cut frequencies are adjusted, I have fixed it for the next build.

 

[John Mulcahy]

Builds updated today with these changes:

• Added: The SPL meter can change its background colour to show when a limit is being approached or has been exceeded
• Changed: Log and linear sweep start frequencies are no longer limited to 20 kHz
• Fixed: Input clipping detection was only active when making sweep measurements
• Fixed: A null pointer exception could occur when opening the RTA if it was in one octave mode and set to show noise curves with adjust RTA levels selected
• Fixed: Incorrect bandwidth could be set when stepping through the octave or 1/3 octave noise centre frequencies

 

[staticV3]

REW can't handle .csv/.txt frequency response files with mixed step size between samples.

I sometimes have to get creative when combining frequency responses, which can lead to some chaotic frequency response files.
For example, here I had to "bridge" two graphs by drawing a line in Illustrator, then tracing that line, then combining everything in a text editor:

green=bridge

The resulting .csv file then looks like this:


Now, software like Plotly or Google Sheets handle this file just fine:

..but not REW:

 

[John Mulcahy]

The code tries to avoid resampling if the data appears to have consistent linear or log spacing that could be retained. For the linear spacing it checked sample intervals at the start, middle and end of the data, which missed the part in that file which had a short period of a different spacing. I'll change it to check the whole file.

 

[staticV3]

The code tries to avoid resampling if the data appears to have consistent linear or log spacing that could be retained. For the linear spacing it checked sample intervals at the start, middle and end of the data, which missed the part in that file which had a short period of a different spacing. I'll change it to check the whole file.

Thank you very much! You rock!

 

[greatadrian]

Hi John, thank you for making REW such an amazing tool! I love the recent additions of importing sweeps whilst preserving relative IR delays, IR overview window and the easy IR export option from "All SPL" window!
I've got a couple of suggestions which I hope you might consider adding in the future.

1. Ability to adjust the gain of the underlying IR.
Right now I can apply offset to SPL but it doesn't modify the measurement. This would come in handy when calculating transfer functions of the audio chain where I can't play the excitation signal at the reference level or if I simply want to match the IR's between the channels to the same calibrated level.

2. Comb filter visualisation in IR view.
Not sure about the UI here, but my initial idea was to add this as part of "Shift+" measure modifier. It would display the peak and notch frequencies, possibly with overlaying them on the frequency response.

3. Using IR as playback equalization
Recently there was a great addition of using EQ during the measurement. I often equalise the speaker for measurements using rePhase and equalizerAPO. These sadly don't work on Mac nor linux. Been thinking whether it's a good idea to use an IR (wav or txt format?) as an equalisation for the playback sweep.

4. Displaying coherence for the measurement
This one is probably just nice to have - we already have THD figures. But calculating coherence between the played back sweep and captured signal could help to quantify how noisy is the channel and the amount of non-linear distortion.

5. Per-channel SPL calibration
Again - this one is nice to have. Right now in multi-mic setup when I perform measurements I am using calibrator on one of the channels to achieve a good target level and calibrate the SPL with external calibrator. This calibration applies to all the channels, so I need to put calibrator on other mics and adjust the input gain so that the reading is the same.

6. Exporting measured sweep
I am not sure if this one is possible as I suspect that measured sweep is not saved as a part of mdat file. But it would be nice to have a way of exporting the recording.

 

[John Mulcahy]

1. Ability to adjust the gain of the underlying IR.
When would this matter? Trace arithmetic takes SPL offsets into account, so shouldn't be an issue for transfer function calculation. It might be useful to have as a feature when exporting the IR as WAV?

2. Comb filter visualisation in IR view.
Responses rarely correspond to expectation for peaks/nulls from reflections as there are so many contributions combining, so I'm doubtful of the utility of this.

3. Using IR as playback equalization
Some shaping of the measurement sweep is on the longer term todo list, but more along the lines of shelving filters rather than arbitrary adjustments.

4. Displaying coherence for the measurement
Coherence is by definition 1 everywhere for a single stimulus and response, it only becomes meaningful when averaging a series of responses such as the processing performed for dual channel FFT measurements.

5. Per-channel SPL calibration
Select "Separate cal for each input" in the mic cal files entry for the input device.

6. Exporting measured sweep
The measured sweep is not retained, keeping it would use a lot of memory.

 

[Lowclock]

Toggling FDW on and off seems to do some weird stuff sometimes, as seen in the group delay of this not-measurement.

Not really sure what is going on with that last one. The first time FDW is turned on it looks the second pic, but every time after that it looks like the third. Info says it switches to 96PPO even though that option isn't enabled?

Edit: Looks like it's something to do with smoothing? 1/48 fixes it for one toggle cycle, anything lower than that sticks and fixes it permanently.

 

[greatadrian]

1. Ability to adjust the gain of the underlying IR
One particular use case for me is when measuring room IR's. First I take the speaker measurement in anechoic chamber at 1m (followed by fixing the level and equalisation). Then I take the setup to a real room to conduct measurements. What I get is a set of IR's that can be compared w.r.t. anechoic measurement at 1m. Then I need to find out what level is needed for anechoic recording to reach 0dBFS across the whole frequency range and apply that gain to room IR's so they act as filters.

2. Comb filter visualisation in IR view
I agree - this is quite specific. I do perform quite a few measurements in anechoic environment so these can be quite visible. It's just a suggestion though

4. Displaying coherence for the measurement
Yes, sorry, I should've added that I mean measurements using repeated sweeps.

5. Per-channel SPL calibration
All these years! Thank you!

 

[Lowclock]

1. Ability to adjust the gain of the underlying IR.
It might be useful to have as a feature when exporting the IR as WAV?

This would be nice. Some convolvers aren't the best at handling things like bandwidth limited subwoofer IRs.

 

[sm52]

First I take the speaker measurement in anechoic chamber at 1m (followed by fixing the level and equalisation). Then I take the setup to a real room to conduct measurements

I'm wondering how an anechoic measurement differs from a normal one. Could you post a measurement of the same speaker in an anechoic chamber and in a real room? It is possible in PM.

 

[John Mulcahy]

Toggling FDW on and off seems to do some weird stuff sometimes, as seen in the group delay of this not-measurement.

Thanks, I have fixed that for the next build. When an FDW has been applied the data will be log spaced.

 

[John Mulcahy]

Displaying coherence for the measurement
Yes, sorry, I should've added that I mean measurements using repeated sweeps.

Doesn't apply there either I'm afraid as the repeated sweeps are synchronously pre-averaged in the time domain ahead of processing.

 

[greatadrian]

Doesn't apply there either I'm afraid as the repeated sweeps are synchronously pre-averaged in the time domain ahead of processing.

I see, makes sense. Thank you John!

 

[John Mulcahy]

Builds updated today with these changes:

• Added: The Impulse graph has a button to scale the IR data to apply a specified gain or achieve a desired peak value
• Added: Impulse import accepts ARTA .pir files
• Fixed: Frequency response import did not cope well with files whose spacing changed for part of the file
• Fixed: Group delay could be incorrect after toggling FDW

 

[Lowclock]

Builds updated today with these changes:

• Added: The Impulse graph has a button to scale the IR data to apply a specified gain or achieve a desired peak value

Works great, except you can't change the value with the mouse wheel like you can with pretty much everything else, which is one of my favorite things about REW!

Also while messing about testing it, I very quickly flicked my free-wheeling mouse wheel on the ETC smoothing value and it just kind of bobbled around calculating between random values between like 0 and 3 and eating up a ton of CPU. Scrolling it again or changing to a different graph didn't stop it, but changing to a different measurement did. Got sick of waiting to see if it would stop after like 15 minutes. Not sure if this affects any other values.

Edit: Tried with some other values and this only seems to happen with ETC smoothing, but I can consistently make it happen with one quick flick regardless of the measurement.
EditEdit: Also, turning off FDW will remove 1/48 smoothing even if it was applied manually before FDW, but it will leave all other forms of smoothing on.

 

[greatadrian]

Added: The Impulse graph has a button to scale the IR data to apply a specified gain or achieve a desired peak value

So quick? This is fantastic, thank you John!

 

[John Mulcahy]

Works great, except you can't change the value with the mouse wheel

I have added mouse wheel support for the next build.

I very quickly flicked my free-wheeling mouse wheel on the ETC smoothing value and it just kind of bobbled around calculating between random values

I can't reproduce that but I have added some mitigations for the next build.

turning off FDW will remove 1/48 smoothing

There isn't a way for REW to know whether the 1/48 smoothing was applied manually or imposed by the 96 PPO log spaced conversion that happens as part of applying the FDW. 1/48 octave smoothing is as close to unsmoothed as 96 PPO data can get, if FDW is removed REW may revert to a linear spaced measurement and then there is a decision about whether to keep the 1/48 octave smoothing or remove smoothing. In recent builds it reverts to unsmoothed, previously it kept 1/48. I can do either and don't mind which, happy to hear what folk think makes most sense.