Go Back   Cockos Incorporated Forums > Projects > Deprecated REAPER issue tracker > Feature Request

LUFS Functions Issue Tools
issueid=3307 04-02-2011 12:04 AM
Human being with feelings
LUFS Functions
Measuring LUFS and LU in Reaper


Updated May 25th 2012 with the Loudness Metering History Graph feature.

Please visit the discussion thread to post your comments.

While we wait patiently for Cockos to implement this, there are several free or inexpensive measurement plugins you can use to at least get basic measurements done and some analysis.

None of these tools have all this request asks for.

ITU-R BS.1770, the basic method to measure loundness, A/85 the document that details its use in North American broadcasting and the EBU R128 recommendation that adds a great deal of practical functionality that is being adopted by the ITU, are the new loudness measuring standard, replacing peak and VU metering. Detailed information can be found via this thread. A video overview of the EBU R128 recommendation is available from the EBU itself on this page.

Where is this loudness measurement used ?

Everywhere where there's digital TV broadcasting.
Europe , US, Japan, Autralia, China and South America.


-edit-
Added an idea on how the loudness graph view could look with elements we already have available in Reaper.
Updated video links to one-part versions just released from the Rome 2011 conference about loudness.


Information Videos
EBU R128 Introduction - Florian Camerer (37:50 for some listening examples, comparing peak and loudness normalization)
Loudness Wars - Thomas Lund
EBU R128 in Transmission and Production - Thomas Lund
EBU R128 in Distribution - Richard van Everdingen
LUFS - the new Loudness Units. What does it mean? by Ian Shepherd


Existing open-source implementations
libebur128 MIT licensed library in ANSI C

https://github.com/jiixyj/libebur128
Discussion thread at Hydrogenaudio
A mix of small sections of music, each individually loudness normalized to -18 LUFS( +5 LU ) - 15 MB
Music provided under Fair Use for educational purpose only(no complete songs, just bits of them). Contact me if you think I unfairly used your music.

Download at Mediafire
Mirror available at Soundcloud here :
https://soundcloud.com/airon-extv/18...lized-clip-mix
Peak overview of that clip mix(-23 LUFS version) here

Because of the significance of this standard to anything broadcast-related in more and more areas of the world(it is already a requirement in North America), I ask that Reaper support at the very least, the R128 standard with the following functions.


Requested Functions

LUFS metering in the master and any other track designated by the user
Momentary(400ms), Short(3s) and Integrated(manual start/stop, only one that includes the gate function). The gate function is now locked at -10 LU. It needs to be on/off togglablefor folks mixing for some North American networks that use -24 LKFS intergrated, which is -24 LUFS without the gate function.

Tracks other than the master need to be included because it is not always just the master that is being measured. Different submix versions need to be measured as well.

Integrated LUFS value of master/selected track(s) over a time selection
Broadcast program material, and in some cases, segments of material between commercials, need to be normalized to -23 LUFS. Thus it will save a great deal of time if the mix or portions thereof can be normalized to that value by calculating the value for a certain period of time and making a manual adjustment.

One way is to render the mix, and then normalize that file to -23 LUFS, which the section below refers to.

However, sometimes it is quicker to simply calculate the LUFS value of a mix or a section thereof and adjust a bus fader to normalize to -23 LUFS before rending to a file, which is what this section of the request would do.

The value would be shown either in a popup requester, or a new track with an empty item that spans the time selection and has the LUFS value printed in the item name, or a text item of equal length, or a region could be created that shows the LUFS or LU value for that time selection.
Loudness History Graph View
Like a peak overview for items, but for the Momentary(and Short-Term if necessary) loudness of the master output, and/or a chosen track/hardware output.

See a detailed description of why and how below.
Normalize item to 0 LU (-23LUFS)
This function can be used to normalize the loudness level of an already rendered mix.

This function could be an action and a button in the item properties.

It also needs to be in the render dialog window as a checkbox and the batch converter.

Why are these functions of value ???

Good question. Here's an example workflow.

Let's sat that I edit and mix a set of 35 x 25 seconds and 5 x 30 second spots that serve as promotional material for a daily drama series on a network, every week.

Each of these spots has to conform to the EBU R128 Loudness spec, just like all segments of program material. That includes commercials, segments of program material sandwidched between commercials, news and so on.

In our example I have 40 separate sequences. But you don't deliver 40 separate files, you deliver one large file that gets layed back to digital video tape(Digibeta,HDCam,DVCPro50, whatever).

What are the options ?
  1. We render each spot to its own file via regions, run those files through an R128 Batch-Normalizing processor and place the resulting files back on the Reaper timeline for consolidation(glue). This takes a lot of time to both prepare the regions, drag the resulting files in to a batch processor, and place each of those files in to the timeline again. Tricks include having the render-regions back to back and numbered correctly so they can be drag'n'dropped in one go.

    All this is error-prone. YOu could forget to name a region properly. You might drag'n'drop all but one file in to the batch processor, or back in to Reaper. Little mistakes could cost you lots of time.


  2. We HAVE EBU R128 normalization in Reaper.

    We render as one large file, then split the result with a mouse-over split macro. There are 30 seconds of dead air inbetween each of those spots, so that will take 10-20 seconds at most.

    All the pieces are selected, EBU R128 normalization of the item gain is performed, and the result is glued together. We're done. We can do our layback with the resulting file and all spots are loudness-normalized.

    Second method for this approach is to region-render everything to one track in the session, normalize all those items to 0 LU, and then glue the suckers in to an item of the correct length and start. Problem solved.

    There are far fewer sources of errors here.

A loudness history graph view in Reaper

Why
Mixing for TV requires you to adhere to specifications of the client, which is often a network like Discovery, CBS or the BBC.

Today these specs are almost aways a form of loudness metering, and more often now, loudness range and Momentary Loudness(the 400ms measurement window) range. So when you fall out of spec, as indicated by a meter with a loudness history graph, you go back to that spot to fix it.

But what if there's more than one spot to fix and the history graph is not locked to the time ruler(/time code) ? What is the meter never knows where it actually is ? You'll have to manually write down all the offending locations, fix them and rescan for confirmation.

What if this could be done more efficiently ?


How

It is built as the user plays back the project.

The history can be updated across a time selection or project wide as needed by manual commands.

For user convenience, Reaper must keep track of the areas of the arrangement that the user has changed or not played yet, and remember that those are the areas it needs to update when the user issues a project or time-selection-based update command.


It could look and work like this



The yellow "Time Selection for rescan.." text section to the right of the yellow'framed time selection wouldnt' appear of course. Maybe the Rescan button would light up if a time selection is made either in the graph window or the arrangement. -edit- Now it does.

There are faint grid lines for the "over" part of the graph, which the graph will then indicate in red as illustrated.

A time selection can be made for rescan or playback. Whether or not the playhead is actually repositioned when you change the time selection is controlled in the preferences. The rescan should be made with a preroll because of delay and reverb effects often being active. The context menu of the window could contain preset values in second increments and a custom setting that brings up a requester. Values should be entered in seconds, which corresponds to what probably all reverb effects use.

Left-click and drag for making a time selection. All keyboard commands applying to zoom and scroll behaviour, which includes the mousewheel-triggered actions, should apply here too without having to painstakingly recreate stuff from the arrangement setup.

No need to repeat the often lamented inconsistencies we had the midi editor. Should a separate context in the action list be made available, perhaps a reference to "As in Arrangement context" for those commands, similar to what some mouse modifier contexts do. In other words, links of one action context to another.


Commands required would be set start and end of time selection, rescan, preroll toggle on/off. Blind folks don't need this view anyway so there really wouldn't need to be commands for creating new views, changing to a different view or removing a view. Removing a view would be handled in the context menu as well, so as to prevent nasty little accidents in the fiddly little tabs.


Configuration
Baseline, maximum momentary LU, maximum LRA. Context menu is the easiest with popups of a combined text entry and dropdown menu(for presets and history).

The baseline (thick green line in the middle) is by default at 0 LU, which equates to -23 LUFS.
The Maximum Momentary LU value can be set to +9 by default. The Maximum LRA is safe at 20.

2nd Update
Updated the image to add a feedback text that appears to the left of the time selection in this view, to indicate the maximum LU, minimum LU and LRA if applicable for that selection. The LRA needs a minimum of 30 seconds iirc to get any meaningful measurements. This gives the user additional feedback and cuts down on guesswork.

The second addition is that little button-box in the top-middle of the selection. This is a function trigger to calculate the integrated LU/LUFS value of that time selection(with preroll if toggled on).

All programs are broken in to segments. Each segment has to conform to the loudness spec of the broadcaster. Thus the mixer will want to check each segment individually without having to render it to a file first. This function was suggested to be an action, but this window is the perfect place to expose that function, and only for time selections as well.

There's nothing else for the user to do on this graph but make a time selection, and in it trigger this useful function. Thus the button won't be in the way at almost any time.


Why this saves time
There are no meters today that lock their loudness history graph to timecode, meaning you can't fix a spot and update the loudness history for it.

This can, and thus nobody has to remember where all the offending locations are because this history graph, locked to the time ruler/time code will always keep you on top of things.

Fix, play or rescan. Confirmed. Next location.


Details and Additional Functions
This is basically a kind of peak overview that should be kept either in the project file or as a separate companion file.

Some folks may wish to have this per-used-hardware-output. 5.1 channel mixes have to be taken in to account as well, though they result in just one graph as well.

Sample frequency is a tough one. Every 0.4 seconds may be a good starting point since that is the integration time of the Momentary Loudness metering anyway. The NuGen loudness meters graph is a good starting point on how to build it. In particular it contains excellent ideas on how to graphically highlight configured thresholds for momentary and short-term loudness measurements.

For example, a spec for a tv network might say :
Integrated Loudness : -23 LUFS / 0 LU
LRA < 15 (loudness range statistically calculated)
Maximum Momentary Loudness: +8 LU
It should be possible to have colourful indicators for when the Momentary Loudness graph crosses that threshold so the mixer can easily locate and fix the parts.

Also, the locations of maximum truepeaks above a configured threshold should be remember as well. This can be important for conforming to specs as well, as some limiters do not perform 4 x oversampled limiting and thus may need to be adjusted. The less time that takes the better, as always.



Issue Details
Issue Type Feature Request
Project Deprecated REAPER issue tracker
Category Audio recording and playback
Status Suggested
Priority 1 - Highest
Suggested Version 3.73
Implemented Version (none)
Users who would use this feature 217
Users who would not use this feature 0
Assigned Users (none)
Tags (none)

05-22-2011 11:57 PM
Human being with feelings
 
Reply
05-23-2011 02:28 PM
Human being with feelings
 
Thanks Dstruct.

As everyone can see, we already have a lot of votes for the pure metering as well, which I'm not tired of telling is nothing but an extra highpass and highshelf filter to produce the k-weighting necessary.

Thanks for voting and spread the word.
Reply
12-08-2012 06:50 AM
Human being with feelings
 
HI I'D LIKE TO SAY THAT -23 LUFS LIITS ARE NOT ACCETABLE

The intent is valid: to put an end to (indeed, hateful) the loudness war, the distortion from wild born "hyper"compressions and so on, for example, to invade homes with commercials or to sell the best dance music.
BUT THERE 'BUT A':
Who makes electronic music or experimental music (also acoustic) has many problems to maintain the loudness at that -23 LUFS level because LUFS due the fact that this parameter also depends on the individual sounds that you got in your mix. So it's not in the mastering stage, but already upstream due the fact that the individual sounds of course inevitably CAN AFFECT THE SPECTRUM OF YOUR AUDIO MIXDOWN: in short terms, if you want to make music intentionally unpleasant and / or from a spectral point of view, "overcrowded" in a good faith, and not necessarily don't need to create a music "harmonious", you will find major limitations with dutiful adjustments that will force you to affect your sonic departure.
In my opinion these limits are -23 ARTISTICALLY UNACCETABLE, as audiophile I have always said that imposing a spectrum sound to an artist means to influence the creativity.
STANDARD ONE MORE 'PROPER AND LESS INVASIVE, in my opinion would have been using the RMS., Also because -in the end- the goal of a wild compression is to increase the RMS.
Setting a maximum RMS level would be guaranteed "freedom spooky" and in some ways, even dynamic range.try it: If you enter in your tracks some starting sound distorted or compressed you are forced to radically alter your mix
Reply
12-08-2012 11:02 AM
Human being with feelings
 
Quote:
Originally Posted by fsuleyman
HI I'D LIKE TO SAY THAT -23 LUFS LIITS ARE NOT ACCETABLE
you are looking to the loudness metering at the wrong way, all that we are talking about in the industry is perceived loudness metering standard recommendations, those are just recommendations for those that could use them to take their art in an even better level.
Reply
12-08-2012 12:10 PM
Human being with feelings
 
Quote:
Originally Posted by Joystick
you are looking to the loudness metering at the wrong way, all that we are talking about in the industry is perceived loudness metering standard recommendations, those are just recommendations for those that could use them to take their art in an even better level.
They are not recommendations in some places and will be mandatory in some countries in a near future.

But they are intended for broadcast only, this has nothing to do with the music industry.

The main idea is just to have consistent levels between programs, channels, networks, etc.
Reply
12-08-2012 12:18 PM
Human being with feelings
 
There is a movement that works for some recommendations in games and other media also, just not quite there yet.

I myself used the recommended measuring standards in music production and is you try you will find a lot of similarities between the ITU and EBU recommendations and the K-System for metering developed by Mr Bob Katz.

In some fields like broadcast they will become standards as part of the CALM act or similar acts. But this has to do with the psycho-acoustically compliance and compatibility between various programs.

All this move is still young and we will see where it leads us in the future.
Reply
12-09-2012 01:44 AM
Human being with feelings
 
many thanks all of you for the nicest replies and infos.
But I remain with my idea : this is a forcing on freedom of expression too excessive.I hope it will be abandoned or fall into disuse.
wish you all nicest dais and good work
Reply
12-09-2012 03:25 AM
Human being with feelings
 
That's just your opinion, and seems like you're in the minority :p
Reply
12-09-2012 03:31 AM
Human being with feelings
 
Quote:
Originally Posted by bendermasteringstudio
But they are intended for broadcast only, this has nothing to do with the music industry.
I see no reason why it shouldn't be done for music too. Listener will benefit from it.
Reply
12-09-2012 07:53 AM
Human being with feelings
 
As I said,except broadcast, those are recommendations and everyone could follow whatever he/she wants. You can use it in the beginning of your mix in order to bring all the vocal takes in the same loudness level, that would be useful, no need to use them only in the mastering stage.

Or you could not use them at all :-)

After all if you try to mix in a reasonable level you will get something very similar to the K-System or the EBU-R128 for Music.
Reply
12-09-2012 08:02 AM
Human being with feelings
 
Quote:
Originally Posted by Dstruct
I see no reason why it shouldn't be done for music too. Listener will benefit from it.
That's a whole other discussion.

The point here is that loudness measurement and regulations are now starting to be a reality in broadcast and because of that Reaper should stay on the game and offer a solution to its users to measure loudness.

That's what is on vote here, nothing to do with music or artistic judgments whatsoever.
Reply
12-10-2012 12:51 AM
Human being with feelings
 
Quote:
Originally Posted by EvilDragon
That's just your opinion, and seems like you're in the minority :p

yes it's just my opinion (as your comment should be a technical opinion ?),
My opinion borns from experience and not from reading the other's opionions in a blog.
It's for who has time to think and reflect a little about it. In the time all will recognize that I m right.
Reply
12-10-2012 06:58 AM
Human being with feelings
 
Quote:
Originally Posted by fsuleyman
STANDARD ONE MORE 'PROPER AND LESS INVASIVE, in my opinion would have been using the RMS., Also because -in the end- the goal of a wild compression is to increase the RMS.
Setting a maximum RMS level would be guaranteed "freedom spooky" and in some ways, even dynamic range.try it: If you enter in your tracks some starting sound distorted or compressed you are forced to radically alter your mix
First of all the r128 specification is using RMS (with a weighting curve) so unless you have a problem specifically with the very simple RLB weighting and/or the high frequency emphasis applied then this specification is very nearly what you recommend.

Second it seems you are suggesting that to standardize and enforce volume levels for delivery is an infringement on creativity. Is that a fair summary? This standard does not in any practical way limit the dynamic range that one can use in their creative works. All it does is attempt to identify one number describing the accumulative loudness over the entire duration of a given work over. Then it uses that number to match the loudness of all works so that one is not on average louder than any of the others.
Reply
12-10-2012 07:40 AM
Human being with feelings
 
thank you so much for your info and x your attention. I agree about what you say but there's a deep difference between RMS and lodness and tha has a deep impact during your mixing. You can measure a low RMS and to read at the same time an in your monitor and "OVER" loudness. As I said the loudness is deeply affected by the single tracks sound in a mixing. Thank you again for your infos. I wish you a good work. nicest regards
Reply
12-11-2012 01:15 PM
Human being with feelings
 
Quote:
Originally Posted by fsuleyman
many thanks all of you for the nicest replies and infos.
But I remain with my idea : this is a forcing on freedom of expression too excessive.I hope it will be abandoned or fall into disuse.
Quote:
Originally Posted by fsuleyman
yes it's just my opinion (as your comment should be a technical opinion ?),
My opinion borns from experience and not from reading the other's opionions in a blog.
It's for who has time to think and reflect a little about it. In the time all will recognize that I m right.
Well, how much experience do you actually have in the field of media law and/or human rights law? Let me just say that my *expert* opinion is that this will in all likelihood not be interpreted as a disproportional restriction on the freedom of expression by consitutional courts. If anything is an excessive infringement on human rights, it is the systematic annoyingly loud advertising that was the primary cause for setting these legal standards. Also, you're in the wrong place and too late to give your opinion, the consulting and discussion phase is over. It's implementation time.
Reply
Reply

Issue Tools
Subscribe to this issue

All times are GMT -7. The time now is 08:41 AM.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2020, vBulletin Solutions Inc.