Old 06-30-2021, 10:22 AM   #1
chmaha
Human being with feelings
 
chmaha's Avatar
 
Join Date: Feb 2021
Posts: 2,261
Default v6.30rc9 - June 30 2021

v6.30rc9 - June 30 2021
  • * Includes feature branch: render normalization
  • * Includes feature branch: EEL2 x86_64/SSE
  • * Includes feature branch: VST3 bridging
  • # JSFX: do not update loudness meter LRA with transport stopped and no live input
  • # JSFX: fix multichannel RMS calculation in loudness meter
  • # Master VU: do not update LRA with transport stopped and no live input
This thread is for pre-release features discussion. Use the Feature Requests forum for other requests.

Changelog - Pre-Releases

Generated by X-Raym's REAPER ChangeLog to BBCode

Last edited by chmaha; 06-30-2021 at 10:29 AM.
chmaha is offline   Reply With Quote
Old 06-30-2021, 12:21 PM   #2
Freiin
Human being with feelings
 
Freiin's Avatar
 
Join Date: May 2020
Posts: 73
Default

Amazing! We're close. Personally, i haven't found any bugs.
__________________
Orchestral composer learning more and more.

https://www.youtube.com/BlackDollMurder
Freiin is offline   Reply With Quote
Old 06-30-2021, 07:09 PM   #3
jrengmusic
Human being with feelings
 
jrengmusic's Avatar
 
Join Date: Jun 2015
Location: Indonesia Raya
Posts: 684
Default

Devs please fix this

annoying web control bug

It also happened in other browsers. It makes the web control unusable.

It’s easy to fix by commenting lines which checks the armed tracks, but it’s not a permanent solution. Because it will be replaced by the annoying buggy version once REAPER updated. If armed tracks should be checked before recording than something wrong with the script.
__________________
JRENG! | M E T R I C
jrengmusic is offline   Reply With Quote
Old 06-30-2021, 08:52 PM   #4
PitchSlap
Human being with feelings
 
PitchSlap's Avatar
 
Join Date: Jan 2008
Location: Vancouver, BC
Posts: 3,792
Default

The new render features are cool.
I think it would better if the font was consistent and there wasn't so much wasted space.

The info button could be in the same row as the others and the loudness statistics section could be ~40% of the current size.

__________________
FRs: v5 Media Explorer Requests, Global Quantization, Session View
Win10 Pro 64-bit, Reaper 6(x64), AMD 3950x, Aorus X570 Master, 64GB DDR4 3600, PowerColor Red Devil 5700XT, EVO 970 2TB, 10TB HD, Define R6
PitchSlap is offline   Reply With Quote
Old 06-30-2021, 10:30 PM   #5
Phazma
Human being with feelings
 
Join Date: Jun 2019
Posts: 2,872
Default

@PitchSlap I agree with you too but in my experience purely aesthetic feedback is rarely considered.

I already pointed out in a previous thread that the font choice looks odd and there was no reaction to that. Any sans-serif font would probably go a long way in looking more comsistent.

Also agree that the buttons should all be in 1 row and the statstics section smaller too. But they probably have reasons for these choices. The big statistics section is probably to incentivize people to scroll if muliple renders/measurements have been made.
Phazma is offline   Reply With Quote
Old 06-30-2021, 10:48 PM   #6
TBProAudio
Human being with feelings
 
TBProAudio's Avatar
 
Join Date: May 2014
Location: Germany
Posts: 643
Default

Quote:
Originally Posted by BethHarmon View Post
v6.30rc9 - June 30 2021
  • # JSFX: do not update loudness meter LRA with transport stopped and no live input

Confirmed.
----------------


First valid LRA value is displayed after 4.8 secs. Should be 3 secs?
__________________
www.tbproaudio.de
TBProAudio is offline   Reply With Quote
Old 07-01-2021, 01:11 AM   #7
n997
Human being with feelings
 
Join Date: Dec 2018
Posts: 503
Default

Quote:
Originally Posted by PitchSlap View Post
I think it would better if the font was consistent and there wasn't so much wasted space.

The info button could be in the same row as the others and the loudness statistics section could be ~40% of the current size.
In my view it is good that results of measurements are in a scrollable field with a different font (as they currently are).

It separates them visually from other UI elements and communicates to user that they are selectable (and, indeed, that there may be multiple rows, as when rendering regions).



As for position of the info button, since the dialog fits on pretty much any display that's likely to be in use nowadays, I personally don't see the extra row of vertical space as an issue.
Having it on its own row right after statistics field seems logical, too, since that's what the info is about.
n997 is offline   Reply With Quote
Old 07-01-2021, 03:22 AM   #8
drew
Mobile
 
drew's Avatar
 
Join Date: Jan 2006
Location: London & São Paulo. Hardcore commercial REAPERite
Posts: 1,669
Default

Just to check, is it by design that if you have a render normalize option on, it doesn't then close the window when finished?

Sorry if that has been covered before, but it does seem to "break" what the checkbox promises.
__________________
Proudly using REAPER exclusively for...
* Media and event music composition & production, sound design + auto-processing at Qsonics.com
* Broadcast branding, promos, education & training and narration voice-overs at DrewWhite.com
drew is offline   Reply With Quote
Old 07-01-2021, 05:34 AM   #9
schwa
Administrator
 
schwa's Avatar
 
Join Date: Mar 2007
Location: NY
Posts: 15,747
Default

The render stats font is the same one REAPER uses for other fixed-width font displays, like in the raw MIDI display for example.

The "automatically close when done" checkbox is not displayed when dry-run rendering or normalizing, because you would typically want to see the results in those scenarios. That's also why there is apparent empty space below the render statistics display, where that checkbox would normally be displayed.

Maybe in a future release we will dynamically resize the render statistics display to be smaller if only one or two files are being rendered, but it's likely to stay as is for the final 6.30 release.

LRA is based on a statistical analysis of 3-second loudness, so it's not displayed until at least 3 seconds have passed, and then enough non-silent audio has been processed to do a meaningful analysis At a minimum, that takes about 5 seconds total.
schwa is offline   Reply With Quote
Old 07-01-2021, 05:53 AM   #10
drew
Mobile
 
drew's Avatar
 
Join Date: Jan 2006
Location: London & São Paulo. Hardcore commercial REAPERite
Posts: 1,669
Default

I understand, thank you. All working great anyway!

I did also test command line with -renderproject and that correctly normalizes then shuts window, so that's fine for unattended use.
(Silent increment doesn't obey checkbox in command line use.. but that may be known and one for another day!)
__________________
Proudly using REAPER exclusively for...
* Media and event music composition & production, sound design + auto-processing at Qsonics.com
* Broadcast branding, promos, education & training and narration voice-overs at DrewWhite.com
drew is offline   Reply With Quote
Old 07-01-2021, 05:59 AM   #11
Zeno
Human being with feelings
 
Zeno's Avatar
 
Join Date: Sep 2018
Location: HH
Posts: 916
Default

Regarding copying the loudness statistics from the render window: Would be great if we could select the text by CTRL+A.

Last edited by Zeno; 07-01-2021 at 08:06 AM.
Zeno is offline   Reply With Quote
Old 07-01-2021, 06:23 AM   #12
Dstruct
Human being with feelings
 
Join Date: Jul 2006
Posts: 12,480
Default

Quote:
Originally Posted by PitchSlap View Post
The new render features are cool.
I think it would better if the font was consistent and there wasn't so much wasted space.

The info button could be in the same row as the others and the loudness statistics section could be ~40% of the current size.

The vertical scrollbar also should be hidden when it's not needed/disabled.
Dstruct is offline   Reply With Quote
Old 07-01-2021, 07:03 AM   #13
Phazma
Human being with feelings
 
Join Date: Jun 2019
Posts: 2,872
Default

Quote:
Originally Posted by schwa View Post
The render stats font is the same one REAPER uses for other fixed-width font displays, like in the raw MIDI display for example.

The "automatically close when done" checkbox is not displayed when dry-run rendering or normalizing, because you would typically want to see the results in those scenarios. That's also why there is apparent empty space below the render statistics display, where that checkbox would normally be displayed.

Maybe in a future release we will dynamically resize the render statistics display to be smaller if only one or two files are being rendered, but it's likely to stay as is for the final 6.30 release.

LRA is based on a statistical analysis of 3-second loudness, so it's not displayed until at least 3 seconds have passed, and then enough non-silent audio has been processed to do a meaningful analysis At a minimum, that takes about 5 seconds total.
Thanks for taking the time to explain the reasons behind all of these things.

I agree, 6.30 shouldn’t be held off for too long, there will always emerge some details and user preferences anyway, but the whole normalization/loudness/render implementation seems solid and stable overall to me. And little improvments can always be made later on in updates (as is happening with RE).

Only suggestion I’d have before release is to put some warning somewhere that normalization won’t work when chosing a video format for rendering. People might expect it, maybe run some time-consuming renders and realize only later that it hasn’t worked or if in a hurry not check and even send out something that doesn’t meet the level/loudness-target.

And our of curiosity, the font choice for these fixed-width displays is hard-coded into Reaper, right?
Phazma is offline   Reply With Quote
Old 07-01-2021, 08:41 AM   #14
chmaha
Human being with feelings
 
chmaha's Avatar
 
Join Date: Feb 2021
Posts: 2,261
Default

Bring on the v6.30 release! The loudness metering and normalization are brilliant additions to an already brilliant DAW. Kudos!
chmaha is offline   Reply With Quote
Old 07-01-2021, 08:56 AM   #15
Klangfarben
Human being with feelings
 
Join Date: Jul 2016
Location: Los Angeles, CA
Posts: 1,701
Default

Quote:
Originally Posted by Phazma View Post
Only suggestion I’d have before release is to put some warning somewhere that normalization won’t work when chosing a video format for rendering. People might expect it, maybe run some time-consuming renders and realize only later that it hasn’t worked or if in a hurry not check and even send out something that doesn’t meet the level/loudness-target.
I agree. As it currently sits it is a bit misleading and a little dangerous as one can check all the normalize options they want to use, choose video as the output format and think they are actually normalizing when they are not.

I also realize it is probably too late in the cycle to consider adding this feature so that you can normalize audio and render to a video output format, but I really hope you consider it. As stated previously, I don't think any of us would have a problem with adding another render pass to accomplish this as any of the current workarounds are much more time consuming.
Klangfarben is offline   Reply With Quote
Old 07-01-2021, 10:10 AM   #16
TBProAudio
Human being with feelings
 
TBProAudio's Avatar
 
Join Date: May 2014
Location: Germany
Posts: 643
Default

Quote:
Originally Posted by schwa View Post
LRA is based on a statistical analysis of 3-second loudness, so it's not displayed until at least 3 seconds have passed, and then enough non-silent audio has been processed to do a meaningful analysis At a minimum, that takes about 5 seconds total.

Basically correct :-)

But if there is a steady non-silent signal then there should be at least a value after 3 secs ... Why wait further 2 secs?
__________________
www.tbproaudio.de

Last edited by TBProAudio; 07-01-2021 at 10:11 AM. Reason: Typo, better english :-)
TBProAudio is offline   Reply With Quote
Old 07-01-2021, 10:41 AM   #17
schwa
Administrator
 
schwa's Avatar
 
Join Date: Mar 2007
Location: NY
Posts: 15,747
Default

Quote:
Originally Posted by TBProAudio View Post
But if there is a steady non-silent signal then there should be at least a value after 3 secs ... Why wait further 2 secs?
The first valid observation is 3 seconds after playback start and subsequent observations occur every 100ms after that. We wait until we have at least 20 non-silent 100ms observations, because the LRA calculation discards the quietest 10% and loudest 5% of observations. If you have say 2 data points, it's not meaningful to discard 5% of them, LRA would be all over the place for the first few seconds.
schwa is offline   Reply With Quote
Old 07-01-2021, 11:20 AM   #18
chmaha
Human being with feelings
 
chmaha's Avatar
 
Join Date: Feb 2021
Posts: 2,261
Default

Quote:
Originally Posted by schwa View Post
The first valid observation is 3 seconds after playback start and subsequent observations occur every 100ms after that. We wait until we have at least 20 non-silent 100ms observations, because the LRA calculation discards the quietest 10% and loudest 5% of observations. If you have say 2 data points, it's not meaningful to discard 5% of them, LRA would be all over the place for the first few seconds.
I can't see anything in EBU-TECH 3342 or ITU-R BS.1770 [2] that suggests there can't be a value returned at exactly 3 seconds if there have already been 20 non-silent 100ms readings (which makes perfect sense given the percentage discarding) as part of the first 3-second window.
chmaha is offline   Reply With Quote
Old 07-01-2021, 11:26 AM   #19
schwa
Administrator
 
schwa's Avatar
 
Join Date: Mar 2007
Location: NY
Posts: 15,747
Default

You have zero readings prior to 3 seconds. After 3 seconds, you have a new reading at least every 100 ms.
schwa is offline   Reply With Quote
Old 07-01-2021, 03:09 PM   #20
chmaha
Human being with feelings
 
chmaha's Avatar
 
Join Date: Feb 2021
Posts: 2,261
Default

Quote:
Originally Posted by schwa View Post
You have zero readings prior to 3 seconds. After 3 seconds, you have a new reading at least every 100 ms.
Could you point me to the place(s) in the spec where it says this? My understanding was that readings were fine during the first 3 seconds but only *reporting* after a complete sliding window.
chmaha is offline   Reply With Quote
Old 07-01-2021, 03:41 PM   #21
schwa
Administrator
 
schwa's Avatar
 
Join Date: Mar 2007
Location: NY
Posts: 15,747
Default

It's a 3 second analysis window. In other words you are analyzing 3 seconds of (overlapping) audio at a time. If you calculate the loudness before you have filled the window with 3 seconds of audio, the calculation will include some silence and be too quiet.

For example if you have media that is at a constant +0dBFS level, it should have LRA of zero. But if you include measurements taken before the analysis window has filled with signal (or for that matter after the end of the material, while the analysis window is emptying), you will get some measurements below +0dBFS and calculate and LRA greater than zero.

You could sort of get around the issue by starting with a shorter window and let it grow to 3 seconds, but you always have to start somewhere. It's in the nature of any calculation that uses overlapping analysis windows that material at the very start and end is relatively underweighted.
schwa is offline   Reply With Quote
Old 07-01-2021, 04:13 PM   #22
chmaha
Human being with feelings
 
chmaha's Avatar
 
Join Date: Feb 2021
Posts: 2,261
Default

Thanks, I think I get what you are saying now!
chmaha is offline   Reply With Quote
Old 07-01-2021, 09:58 PM   #23
TBProAudio
Human being with feelings
 
TBProAudio's Avatar
 
Join Date: May 2014
Location: Germany
Posts: 643
Default

Quote:
Originally Posted by schwa View Post
The first valid observation is 3 seconds after playback start and subsequent observations occur every 100ms after that. We wait until we have at least 20 non-silent 100ms observations, because the LRA calculation discards the quietest 10% and loudest 5% of observations. If you have say 2 data points, it's not meaningful to discard 5% of them, LRA would be all over the place for the first few seconds.
Sound logically, but I cannot find this in the specs.
The only thing I can find is this (regarding meaningful measurement):

tech3341, 2.4 Loudness Range measure

Quote:
... During the first 60 s of the measurement of LRA (e.g. after ‘reset’) the meter shall indicate that the displayed LRA value is not yet to be considered stable. The individual implementation of the meter may determine the means of providing this indication. ...
At least good to know how LRA calc is done :-)
__________________
www.tbproaudio.de
TBProAudio is offline   Reply With Quote
Old 07-02-2021, 12:34 AM   #24
chmaha
Human being with feelings
 
chmaha's Avatar
 
Join Date: Feb 2021
Posts: 2,261
Default

Quote:
Originally Posted by TBProAudio View Post
Sound logically, but I cannot find this in the specs.
What I did find in the spec is that there's a 3-second sliding analysis window so the first report/reading can only happen after 3 seconds. Then, if I'm understanding schwa correctly, you can't remove the required loudest 5% if you only have one reading...mathematically you need 20 such readings before 5% can equal an integer value and be in any way meaningful.

It sounds like a lot has to be inferred from the various specs which is never ideal. Good job they have the test files

Last edited by chmaha; 07-02-2021 at 12:55 AM.
chmaha is offline   Reply With Quote
Old 07-02-2021, 04:11 AM   #25
schwa
Administrator
 
schwa's Avatar
 
Join Date: Mar 2007
Location: NY
Posts: 15,747
Default

There is quite a bit that needs to be inferred from the EBU loudness specifications. As has been commented before, it's an odd combination of very technically specific and surprisingly vague.
schwa is offline   Reply With Quote
Old 07-02-2021, 04:22 AM   #26
Phazma
Human being with feelings
 
Join Date: Jun 2019
Posts: 2,872
Default

Quote:
Originally Posted by schwa View Post
There is quite a bit that needs to be inferred from the EBU loudness specifications. As has been commented before, it's an odd combination of very technically specific and surprisingly vague.
Thanks for having decided to go through all the craze with LUFS implementation after we kept requesting it in the first few dev cycles. I think most of us underestimated the challenge, but you managed to implement the calculation and normalization with great precision. The hard work is very much appreciated!
Phazma is offline   Reply With Quote
Old 07-03-2021, 03:47 AM   #27
PitchSlap
Human being with feelings
 
PitchSlap's Avatar
 
Join Date: Jan 2008
Location: Vancouver, BC
Posts: 3,792
Default

Quote:
Originally Posted by schwa View Post
The render stats font is the same one REAPER uses for other fixed-width font displays, like in the raw MIDI display for example.

The "automatically close when done" checkbox is not displayed when dry-run rendering or normalizing, because you would typically want to see the results in those scenarios. That's also why there is apparent empty space below the render statistics display, where that checkbox would normally be displayed.

Maybe in a future release we will dynamically resize the render statistics display to be smaller if only one or two files are being rendered, but it's likely to stay as is for the final 6.30 release.

LRA is based on a statistical analysis of 3-second loudness, so it's not displayed until at least 3 seconds have passed, and then enough non-silent audio has been processed to do a meaningful analysis At a minimum, that takes about 5 seconds total.
That makes sense. Dynamic resizing is what I'm getting at. If the majority of rendering is of a single file the default size should be for that. Making the whole render window resizable would be ideal, I've often found the waveforms can look a bit squished.
__________________
FRs: v5 Media Explorer Requests, Global Quantization, Session View
Win10 Pro 64-bit, Reaper 6(x64), AMD 3950x, Aorus X570 Master, 64GB DDR4 3600, PowerColor Red Devil 5700XT, EVO 970 2TB, 10TB HD, Define R6
PitchSlap is offline   Reply With Quote
Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT -7. The time now is 04:40 AM.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2024, vBulletin Solutions Inc.