REW Beta Release REW API beta releases

[serko70]

Like the one added in beta 6?

That swaps places with the measurement at the top if I remember correctly. I usually need other measurements not moved but my cases might be too specific.

 

[John Mulcahy]

Builds updated today (beta 15) with these changes:

• Added: A GET at the /import/impulse-response endpoint returns the path of the last impulse response file imported
• Added: A GET at the /import/frequency-response endpoint returns the path of the last frequency response file imported
• Added: /measurements API MeasurementSummary now includes the SPL offset, align SPL offset, clock rate adjustment and cumulative IR shift
• Added: The tone burst signal can continue at a defined "off" level (which is a percentage of the burst level) between bursts if the period is in cycles
• Changed: When importing multiple files sort the list by filename
• Changed: Revised Actions panel layouts to reduce width
• Changed: Maximum measurements increased to 300
• Fixed: After a target match finished all filters were set to Auto

 

[powerplaymike]

Builds in the first post updated today (beta 5) with these changes:

• Added: Action to Change cal for all selected measurements
• Added: Equaliser entries for miniDSP Flex and Flex with Dirac Live
• Added: Graph capture dialog has a selector for the colour scheme of the captured image
• Added: Subscriptions can be added to the /rta/status endpoint to be notified of the RTA starting and stopping
• Changed: The API RTAStatus object includes a boolean to indicate whether the RTA is running
• Changed: The oldest supported macOS version is now 10.15 as Java runtimes are no longer built for older macOS versions

My apologies as I tried a search on the forum.....I always struggle with finding the "correct" place to post a question about the current build, but anyway:

Re: Equaliser for Flex, was hoping to get best course of action for the following:

Was previously running a 2x4HD with filters created using (I think) the 2x4HD equaliser option or whatever was most recent prior to this new build (I was running a beta but don't remember which one, but I'm pretty sure it had 2x4hd as option but not the flex option yet)

Anyway, about 4 months ago, upgraded to an analog/unbalanced Flex to replace the 2x4HD, and imported the same filters (since the Flex option wasn't built yet.)

Saw now that one of the most recent updates now has a Flex equaliser setting as of a month or two ago

Do I need to re-do filters or anything? Or can I just keep with the 2x4HD filters previously used?
Rather not have to break out all the equipment again if I don't need to

 

[John Mulcahy]

Do I need to re-do filters or anything?

No, the new option doesn't do anything differently.

 

[Cristianolo]

is it possible to add more measurements as channels in Alignment Tool, to have a total like 4 ou 5?

 

[John Mulcahy]

That would be difficult to manage, but it is quite easy to switch between measurements on the alignment tool.

 

[John Mulcahy]

Builds updated today (beta 16). There is just one change, but hopefully an interesting one:

• Added: Match target has options to use shelf filters to help bring the response to the target

The filter tasks pane has gained new controls to select whether shelf filters may be used at the beginning (low) and/or end (high) of the response. Shelf filters will only be used if the equaliser supports them and the response remains above or below the target for a span of at least half an octave at the beginning or end of the match range. The allowable gain range for the shelf filters is set to the right of the check boxes.

 

[sm52]

John, is it possible to make this change in the Alignment Tool? Area 5 in the picture is highlighted with turbidity. When working with sliders, you can highlight areas 1, 2, 3, 4 with this blurring. And leave areas 5, 6 normally visible. In area 5, when selecting delays, something changes in the black graph, in the resulting graph, because the levels of the original graphs range from 0 to -30 (approximately) dB from the main one. When added, they have a noticeable effect on the result. In areas 1 and 3, nothing changes because the levels of the lower graphs are too low. It is difficult to understand up to what frequency to pay attention to the addition of phase and amplitude graphs, and where it is not important. Perhaps you can come up with a way to highlight the graphs in areas 5 and 6 by highlighting. But it seems to me that highlighted with turbidity areas 1, 2, 3, 4 is better. You can also give the user the opportunity to specify at what difference in levels to draw the line between significant and insignificant areas.

 

[John Mulcahy]

John, is it possible to make this change in the Alignment Tool?

Wouldn't it be easier to just zoom into the frequency range you are interested in?

 

[serko70]

Dear John,

Server starts responding incorrectly after repeated measurement delete requests. There're no other responses than this and 200 so, it's not possible to catch that problem, either.

 

[serko70]

Builds updated today (beta 16). There is just one change, but hopefully an interesting one:

• Added: Match target has options to use shelf filters to help bring the response to the target

The filter tasks pane has gained new controls to select whether shelf filters may be used at the beginning (low) and/or end (high) of the response. Shelf filters will only be used if the equaliser supports them and the response remains above or below the target for a span of at least half an octave at the beginning or end of the match range. The allowable gain range for the shelf filters is set to the right of the check boxes.

View attachment 68894

Is there a way to disable peaking filters and force use of only HS for the higher frequencies?

 

[John Mulcahy]

Is there a way to disable peaking filters and force use of only HS for the higher frequencies?

No.

 

[John Mulcahy]

Server starts responding incorrectly after repeated measurement delete requests.

The API only responds to requests, so your script must be making the requests, must it not?

 

[serko70]

The API only responds to requests, so your script must be making the requests, must it not?

Apparently, I can slow down the requests and can delete them all but receiving "Measurement X is deleted" response from the server when in fact measurement X is not deleted and is not even in the server's pipeline is a bit odd. Every other post API returns some sort of completion message which are fully reliable. Delete is the only one with this strange behavior..

 

[serko70]

The code below will iterate through as many posts as you send without logging a single "Network response was not OK" or an "Error fetching result" or applying any delay as it will not ever go to the promise resolve line, either, Because the response is always 200 and the message is always "Message x deleted" no matter how fast you send requests. Unless you slow down calls which has varying reliability depending on the machine it runs, you will risk being unaware of what is really deleted: I tried matching deletion message indices before sending the new request as well but didn't help!

 

[John Mulcahy]

Builds updated today (beta 17) with these changes:

• Added: The Predicted SPL overlay graph has an action to apply the equaliser and filter settings from one measurement to all other selected measurements
• Changed: Allow IEC263 formatting of distortion graphs if the Y axis is SPL, dBFS or dBr
• Changed: Allow IEC263 formatting of the decay graph
• Fixed: Excessive filter use in match target when individual boost filters were allowed
• Fixed: API could respond OK to a measurement deletion if one was already in progress

 

[serko70]

Builds updated today (beta 17) with these changes:

• Added: The Predicted SPL overlay graph has an action to apply the equaliser and filter settings from one measurement to all other selected measurements
• Changed: Allow IEC263 formatting of distortion graphs if the Y axis is SPL, dBFS or dBr
• Changed: Allow IEC263 formatting of the decay graph
• Fixed: Excessive filter use in match target when individual boost filters were allowed
• Fixed: API could respond OK to a measurement deletion if one was already in progress

John, a +gain LS filter seems to increase C80 quite significantly in the bass region compared to just peaking filters. I know C80 is not very meaningful in small rooms but it's still the only one that sort of captures sound energy and is usually good sign of a sound throttling filter IMO. I wonder how is applying boost in bass frequencies can have such positive effect. There seems to be a very special algo working behind the scenes?

 

[John Mulcahy]

I know C80 is not very meaningful in small rooms

Or not at all meaningful, particularly for low frequencies. The low shelf increase the low frequency content of the impulse response, so there is more energy in the response before it decays into the noise. A low shelf operates over all frequencies below its centre frequency so it affects a wider bandwidth than a PK filter.

 

[serko70]

Index out of bounds exception during delete fetch, admittedly I forgot to use await at the beginning so it was an absurd fetch request but you wanted any java exception to be shared, so:


REW V5.40 Beta 17 running Azul Systems, Inc. JRE 1.8.0_402 64-bit on Windows 10 Locale: language en, keyboard GB, windows-1252 at 96 DPI Running in C:\Program Files\REW


Message:
java.lang.ArrayIndexOutOfBoundsException: 111
Stack Trace:
111
roomeqwizard.QI$_B.run(y:1830)

 

[powerplaymike]

No, the new option doesn't do anything differently.

Great thanks John!

 

[John Mulcahy]

Builds updated today (beta 18) with these changes:

• Changed: Keep audio capture running during repeated and sequential measurements rather than stopping and restarting it
• Changed: Inversion arithmetic operations have a maximum gain setting instead of a regularisation figure
• Changed: Invert phase does not need a regularisation figure
• Changed: Improved regularisation/gain limiting and Exclude notches behaviour for inversions
• Changed: The minimum phase copy of a measurement keeps the window types and widths of the source measurement
• Fixed: Division with regularisation did not work properly when one of the measurements did not have an IR
• Fixed: Inversion used the unsmoothed response

 

[Lowclock]

E: I need to investigate some more.

 

[John Mulcahy]

Builds updated today (beta 19) with these changes:

• Fixed: Inversion exclude notches option did not work well

 

[Lowclock]

Dangit John I didn't even get to finish a coherent post about it and you already fixed it! LOL Thanks again. You're the best.

 

[Lowclock]

Actually, should these dips around 415 and 600hz really not be considered notches? They have a pretty sizeable excess group delay spike, and behaves like ones in practice. The old notch logic excluded them but not the new one.