Trying to understand Reaper's stereo balance algorithm

[DrTT]

Dear all,

I am fiddling around with reproducing Reaper's signal manipulation behavior in an external audio chain.

I am quite successful there, because Reaper behaves logically, but one open issue is how Reaper handles stereo panning. I tried to measure this by taking a simple test file in a track with a 1kHz stereo square wave on one track, rendering different pan positions in Reaper to external files and analyzing them.

The following points came up during my analysis:

• Even for a symmetrical signal without any DC-offset sometimes the stereo-balanced signal has some slight DC-offset. It is minimal (about 0.01%), but it exists.
  In my opinion panning shouldn't add a DC-offset for a symmetrical signal.
• I am using the default pan law, but nevertheless there seems to be some unclear amplification of the signals.
  The enclosed diagram shows the amplification factors when turning the balance knob. At "center" both signals are amplified at 100%, then the main signal (on the panned side) increases and decreases while the signal on the panned-away side linearly decreases.
  The latter is plausible; it does not look perfectly linear (the points should be at raster positions), but this seems okay to me. Unfortunately for the panned side my naive assumption is that the amplification factor should be constant. But the amplification at 100% pan is even 4% higher than for 0% pan.

A possible explanation is that panning/balancing is emulating some hardware pan pots with that behavior or some kind of loudness compensation is done, but this is just a guess.

I am pretty puzzled, so can anybody help me out here or - even better - give me the math behind that mechanism? It might also be possible that I have some systematical measurement error in my approach, so I am grateful for any hint.

Best regards,
DrTT

[schwa]

Probably everything you ever wanted to know about pan law in REAPER:
https://forums.cockos.com/showthread.php?t=231241

[Thonex]

Cockos has done various changes since that thread... DC offset aside, I think the conclusion is this:

Originally Posted by Sekkleman

I decided to try and figure out what's going on with the Reaper pan law options as I need to be certain my mixes will translate between Reaper/Pro Tools/Cubase without any issues.

Here are my findings which may be of interest to those that are going between these DAWs -

The source of the confusion with the Reaper pan laws comes down to the fact that Reaper doesn't have real mono tracks, only stereo. In the other DAWs I tested (Cubase/Pro Tools) the pan law is only implemented on mono tracks and not on stereo tracks. Therefore the fact Reaper has a pan law for it's stereo tracks doesn't line up.

I tested a mono sine wave in Reaper, Cubase and Pro Tools with a 3db pan law and they all acted the same - center position is 3db lower than hard left or right. This makes sense as the 3db pan law means a mono signal is attenuated when panned to the center to compensate for it coming out of both channels of the stereo buss it's routed to.

For a stereo track in Reaper to act the same as a mono track in Cubase/Pro Tools with a default 3db pan law (or equal power), it needs to be set to -

1. -3.00db
2. Pan mode - Stereo balance / mono pan

As mentioned before - the projects pan law setting is not implemented on stereo tracks in Cubase or Pro Tools. Cubase uses a 'Stereo Balance Panner' or 'Stereo Combined Panner', not the pan law, while Pro Tools uses a 'Dual Mono Panner (same as the Cubase Stereo Combined Panner) for the left/right signals and not the pan law. This is why Reaper having a pan law on it's only track type (stereo) has never worked as expected when coming from these other DAWs, as pan laws are used only for mono tracks with mono signals going into stereo busses.

For a stereo track/buss in Reaper to act like a Stereo Balance Panner for a stereo track/buss in Cubase it needs to be set to -

1. 0.00db
2. 'linear scale above 3db' ticked
3. Pan mode - stereo balance

For a stereo track/buss in Reaper to act like a Stereo Combined Panner for a stereo track/buss in Cubase it needs to be set to -

1. -3.00db
2. 'linear scale above 3db' ticked
3. Pan mode - dual pan

For a stereo track/buss in Reaper to act like the Dual Mono Panner on stereo tracks/busses in Pro Tools it needs to be set to -

1. -3.00db
2. 'linear scale above 3db' ticked
3. Pan mode - dual pan

If all tracks that are supposed to act like mono tracks in Cubase/Pro Tools are set in the above configuration, and all tracks that are supposed to act like stereo tracks/busses in Cubase/Pro Tools are also set in the above configurations, the panning/gain staging should translate correctly between the DAWs.

There also isn't a need to use the 'Gain compensation (boost pans)' setting either, as a panned 'mono' Reaper track will always be routed into a 'stereo' Reaper buss which will therefore compensate for this (obviously there's no point routing a true mono track to another true mono track as it would defeat the pan). This setting kind of helped fix the issue of confusing 'stereo' configured tracks with 'mono' configured tracks and the subsequent gain loss when bussing them to eachother. So in the end, having a default project pan law that affects both perceived mono tracks (where you have mono audio on a stereo track) and the actual stereo tracks in Reaper won't give you the same pan law or gainstaging in a traditional sense.

I think the simplest way around this is to create track templates for 'mono tracks' and the two types of 'stereo tracks'. This will allow anyone to easily build stereo mix sessions while ensuring mix compatibility with Cubase/Pro Tools.

[DrTT]

Hello Schwa and Thonex,

thanks for your hints towards the discussion threads!

The main message from those threads seems to be that the pan law is applied nonetheless to "stereo tracks" and hence this distorts the amplification for the panned-to channel.

Unfortunately this makes it hard to emulate that behavior externally, so I am resorting to writing a simple JSFX-plugin that behaves in the naive way I like. It's a pity that then I cannot use the standard controls in the control panels, but I can live with that.

Nevertheless the non-linearity of the panned-from channel and the DC-offset should be fixed some time. But I have to admit that I have a very academic and measurement-oriented approach to music production, hence this might not be time-critical or even relevant to others.

Best regards,
DrTT

[norbury brook]

Originally Posted by Thonex

Cockos has done various changes since that thread... DC offset aside, I think the conclusion is this:

hey T,

How does this play out with folders/busses?

I remember back in the day I used to set to 3db pan like in cubase but something bout folder tracks meant things were 3db down when you did this.

Is this still the same.

At the end of the day if you use your ears does it matter? it only matters realy if you want to translate a mix from reaper to another DAW or vice versa no?

M

[DrTT]

Dear all,

I should be reading the descriptions more carefully: the behavior I like to have is already implemented in the "old" Reaper 3.x balance mode: it keeps the panned-to channel at an amplification of 1 and linearly decreases the panned-from channel.

Unfortunately this selection is tagged as "deprecated", but I hope this does not mean "will be removed in a future release".

Thanks for all the hints!

Best regards,
DrTT

[schwa]

We are in fact planning to remove the deprecated, linear pan mode from future versions, although existing projects that have the pan mode set that way will not change. Linear panning is really not appropriate except maybe in some very technical applications. This is the stereo image that results:



There is a reasonable discussion of why at the bottom of this page:
https://www.cs.cmu.edu/~music/icm-on...dings/panlaws/

Plenty of other online resources are available on this topic as well.

[norbury brook]

@Schwa


what do you suggest we set our projects to? I always used -3db pan LAW or perhaps 4.5 but I remember from years ago this messed up folder tracks in reaper.


M

[schwa]

There's a pan mode, and a pan law. The default pan mode since version 4.0 has been stereo balance / mono pan; the default pan law has been +0dB.

My previous comment was about the pan mode. There is a deprecated linear mode (deprecated in 2011) that we may remove in future versions.

Re pan law, the main reason REAPER's default is +0dB is because it gives predictable output when used with folder tracks. If you don't care about that, then any pan law is fine. I think the default for most DAWs is -3dB or -4.5dB.

[norbury brook]

Originally Posted by schwa

There's a pan mode, and a pan law. The default pan mode since version 4.0 has been stereo balance / mono pan; the default pan law has been +0dB.

My previous comment was about the pan mode. There is a deprecated linear mode (deprecated in 2011) that we may remove in future versions.

Re pan law, the main reason REAPER's default is +0dB is because it gives predictable output when individually panning multiple child tracks within a folder track. If you are not doing that, then any pan law is fine. I think the default for most DAWs is -3dB or -4.5dB.

yes, that was the issue, panning child track inside a folder messed things up.

If a folder is thought of as a simple Aux/Bus track, why does the main law thing mess this up? other DAW's don't have this issue no?

So, for example if I use a regular track and make it a bus and send ting to it, doe's the pan law work as expected then? if so why does a folder mess this up?


M

[schwa]

The issue is the same with a stereo send. If you have a stereo track with identical left and right channels and send the stereo signal to another track, whether as a folder child or as a regular send, the stereo output level of the receiving track will be lower by the amount of that track's pan law. (Note that sends have their own pan law as well; if the send pan law and the receiving track pan law are both -3dB, then the receiving track output will be -6dB compared to the sending track.)

It's not about right or wrong, it's about predictability. If you use folder tracks for organization, you would not expect the output to drop just because you put a track in a folder. If you are not bothered by the level change, then by all means use a different pan law.

To be clear, this is just how stereo summing and pan law works in general, it doesn't have anything to do with REAPER specifically.

[norbury brook]

Originally Posted by schwa

The issue is the same with a stereo send. If you have a stereo track with identical left and right channels and send the stereo signal to another track, whether as a folder child or as a regular send, the stereo output level of the receiving track will be lower by the amount of that track's pan law. (Note that sends have their own pan law as well; if the send pan law and the receiving track pan law are both -3dB, then the receiving track output will be -6dB compared to the sending track.)

It's not about right or wrong, it's about predictability. If you use folder tracks for organization, you would not expect the output to drop just because you put a track in a folder. If you are not bothered by the level change, then by all means use a different pan law.

To be clear, this is just how stereo summing and pan law works in general, it doesn't have anything to do with REAPER specifically.

thanks @schwa , but if I'm using pro tools or cubase, sending my tracks to a bus doesn't have any effect on the track pan levels the project pan law stays the same.

my point really is/was why does reaper change this paradigm ?

To me as an old school console guy , I don't have to think about bussing tracks to a bus and wondering if the level changers in the pan law.


Is this ONLY an issue if transferring between DAW's other than Reaper?


M

[ernzo]

So what's the difference between the "Reaper 3.x balance (deprecated)" Linear mode,
and the regular "Stereo Balance/Mono Pan" set at +0db?

[schwa]

I'm probably going to have to bail out of this thread because pan law is a real energy vampire, but to answer that last question, "stereo balance, +0dB pan law, linear scale above -3dB" is equivalent to "3.x deprecated, +0dB pan law". But "stereo balance, -XdB pan law, linear scale above -3dB" is not equivalent to "3.x deprecated, -XdB pan law".

Here is a project set up to meter 4 different combinations of pan mode and pan law, using only JSFX. You can probably answer any questions you have by playing around with it.

[ashcat_lt]

Originally Posted by norbury brook

my point really is/was why does reaper change this paradigm ?

It’s basically because of the universal track thing. In other DAWs, a track is a track and a bus or aux is a completely different thing. Reaper itself makes no such distinction. They’re all just tracks. You can route any track to any other track, and you can call one of them a “bus” or an “aux” if you want, but it doesn’t change anything about the way Reaper sees or treats it.

That is, Reaper has no good way of knowing which tracks you want pan law on and which you might not unless you specifically go in and override the pan law at the track. This is a thing I don’t think other DAWs even let you do. In most, I think, if you even can change pan law, it would be either global or by project (session, whatever they call it) with no way to change it for just one or a few tracks. If it really bothers you the way Reaper does it, you can set up different track templates with different pan laws for the different uses.

[mlprod]

If you want a pan law just use "boost pans" and you'll always have unity when pan is center. I agree with schwa that there's way too much pan law talk in forums...

[norbury brook]

Originally Posted by mlprod

If you want a pan law just use "boost pans" and you'll always have unity when pan is center. I agree with schwa that there's way too much pan law talk in forums...

I think the issue here though is, in all other DAW's you set your pan law globally and that's it.

In reaper if you do this by setting say a 3 db pan law, when you send to a folder you'll find your panned child tracks are not the same level they would be before they were sent to the folder, hence the default 0 db pan law being recommended in reaper.

this is the downside of NOT having defined stereo or mono tracks.

M

[mlprod]

Originally Posted by norbury brook

In reaper if you do this by setting say a 3 db pan law, when you send to a folder you'll find your panned child tracks are not the same level they would be before they were sent to the folder, hence the default 0 db pan law being recommended in reaper.
M

This would of course only happen if the folder parent is also panned. Mine are never panned. But yea I never mess with this, I made a decision years ago and stick with that.

[ashcat_lt]

It’s not about panned children. In the normal non-gain compensated modes, both channels are attenuated when the pan is centered. Since folder parents, buses, and even auxes are rarely actually panned, you just end up losing the pan law’s worth of level no matter what you send them.

I personally just use 0db for everything and quite naturally compensate for level changes with the fader if/when I feel it’s necessary. The only time pan law is really warranted is when automating pans, and frankly even then, I’m not often bothered by it. Any plugin that does panning is going to have its own pan law, and very often if I’m panning something toward the center, I also do want it to step forward a bit anyway.

[Mercado_Negro]

My workflow is:

For tracks with mono output use -3dB as pan law and the default mode

For tracks with stereo output use -3dB as pan law and change to dual pan

I only apply this if I need to use automation on that track otherwise I just leave it be.

[Klangfarben]

Originally Posted by Mercado_Negro

My workflow is:

For tracks with mono output use -3dB as pan law and the default mode

For tracks with stereo output use -3dB as pan law and change to dual pan

I only apply this if I need to use automation on that track otherwise I just leave it be.

The problem here and that norbury already touched on, is that there isn't really a way to make one setting work for both mono and stereo tracks, whereas this is not an issue in other DAWs due to having distinct mono/stereo tracks.

I'm certainly not trying to make the argument Reaper should adopt this approach, but I think the argument can be made that this is much more convoluted than it should be, especially for new users. This should be pretty simple but in its current state, it just isn't.

Maybe it is time to consider different pan settings for mono/stereo items in the project settings? That way there would be an "auto" way to set and forget based on if a track has only mono items or only stereo items. Not sure about tracks that mix and match though, but I also can't say I would ever, ever do that so I don't think it would really affect the users who are doing that.

Also, more than a few people now have stated that the 3.x deprecated setting does what they expect. I'm not sure it is a good idea to remove it. In fact, I think it should be renamed so that users who are struggling can find it a little easier when they are trying to find the right setting as generally they aren't going to try something called "deprecated".

Just my inflation-based 3 cents but I really think this needs to be less complicated and easier to navigate than it currently is.

[Klangfarben]

Originally Posted by schwa

Linear panning is really not appropriate except maybe in some very technical applications.

We really have three ways to accomplish pan law. Linear, square law and sine law. But honestly, that's not what is important here. What is important is equal gain and equal power. Linear results in equal gain/amplitude. Square and sine law result in equal power.

As others have pointed out, Cubase/Nuendo, Logic and Pro Tools all use equal gain for their pan law. So saying that linear panning is not appropriate except in some technical applications is NOT accurate at all because most DAW users whether they realize it or not are using equal gain for pan law, not equal power.

It also means that Reaper users coming from any of those DAWs or receiving files from any of those DAWs are going to have to jump through hoops just to get the pan law sounding the same/similar.

Originally Posted by schwa

There is a reasonable discussion of why at the bottom of this page:
https://www.cs.cmu.edu/~music/icm-on...dings/panlaws/

There are two issues with this article. First, is that while the math is correct, the opinion of which approach is better is simply that of one person. Second, even if you agree with this person's opinion, the reality again is that most DAW users are in fact using equal gain, not equal power because that is how the DAW they are working in is coded. The author also doesn't take into account -6dB pan law as an option at all, even though they really should have.

So, when you say, bah, linear, no one uses it or should use it, let's bail on it, that's pretty concerning because you are imposing that choice on everyone. Now, you are the captains of the ship, so imposing your choice on us is completely captains prerogative. But in this case to be fair, your reasoning for doing so is a bit suspect and is making this unnecessarily complicated, especially for new users.

Think of it this way. If you are doing complex editing work, sometimes an equal power fade works when an equal gain fade doesn't and vice versa. You are never exclusively using one or the other. It is the same situation here. Sometimes people are going to want an equal gain pan law and sometimes they aren't.

As you can see from the people coming from other DAWs who are very confused, it is because they are used to using equal gain while Reaper defaults to equal power. And just going through the options in the project settings, it's really hard for them to figure out which one they want to be using. I think most users are going to have no idea what a "Linear scale above -3dB pan law" checkbox means whereas if you rename the 3.x deprecated option to "equal gain pan law" or something similar that is less confusing. Again, when multiple users are saying they couldn't recreate the correct pan law for them until they selected the deprecated option, imho you shouldn't get rid of the deprecated option.

[schwa]

The deprecated mode is only equal gain at +0 or -6 pan law. In between it looks like this. It was deprecated for a reason.



The default mode is equal gain at +0dB if "linear scale above -3dB" is checked, and linear at -6dB by definition.

I suppose... what we could do is deprecate both the 3.x mode and the "linear scale above -3dB" checkbox, and instead have an equal power vs equal gain control. Equal gain at other than +0 or -6 dB is not currently available in REAPER.

[Klangfarben]

Originally Posted by schwa

I suppose... what we could do is deprecate both the 3.x mode and the "linear scale above -3dB" checkbox, and instead have an equal power vs equal gain control.

I think this is an excellent idea and greatly simplifies everything.



It's basically the same options, but allows users to better select the available choices without any confusion. And for new users who are confused and asking on the forum, someone can just tell them, select equal gain instead of equal power (not that these threads aren't informative )

Also, for clarity's sake, you might want to consider removing the -6.02 option. Yes, that is exact value of the combined gain of the channels, but it exists only on paper. It doesn't exist in the real world which is why everyone just uses -6dB instead. Having BOTH -6dB AND -6.02dB is confusing and a bit unnecessary. IMHO it would be better just to have the list like this:



There's no confusion there and simplifies things for everyone.

Originally Posted by schwa

Equal gain at other than +0 or -6 dB is not currently available in REAPER.

As far as equal gain for the other values, again captain's prerogative. But I would say this. It would make it so much easier for new users migrating to Reaper which is what we want, right? Also, since you've made the correction and the math is already done, I'm guessing it would not be difficult to implement?

**and thank you for the clarification on the deprecated option. I completely forgot about that from the other thread.

[schwa]

pan law changes are in prerelease builds now.

https://forum.cockos.com/forumdisplay.php?f=37

[Thonex]

Thanks so much, Schwa and Devs!!!

I'm sure this will be a great update to the old Pan law functionality.

Thanks again for doing this.

Cheers,

Andrew K

[norbury brook]

Originally Posted by Thonex

Thanks so much, Schwa and Devs!!!

I'm sure this will be a great update to the old Pan law functionality.

Thanks again for doing this.

Cheers,

Andrew K

+1 as there's no 'like' button on this forum.....

feature request?


M

[thevisi0nary]

This is a subject I've had a frustrating time understanding in the past, especially when it comes to things like folders.

To ask a very amateur and simple question, does this have bearing on ANYTHING other than the volume of a signal when panned? Does it affect phase /frequency/my vsti's/anything aside from strictly volume?

[ashcat_lt]

Originally Posted by thevisi0nary

…does this have bearing on ANYTHING other than the volume of a signal when panned?

Well, depending on the law, it also affects volume when centered, but yeah it’s only about signal levels. If it changed other things we’d have much bigger problems.

[thevisi0nary]

Originally Posted by ashcat_lt

Well, depending on the law, it also affects volume when centered, but yeah it’s only about signal levels. If it changed other things we’d have much bigger problems.

Okay got it, thank you much. As long as I only have to worry about balancing levels when panning then I don't have to overthink this anymore lol.

[norbury brook]

Originally Posted by thevisi0nary

Okay got it, thank you much. As long as I only have to worry about balancing levels when panning then I don't have to overthink this anymore lol.

exactly, the issue with Reaper was though that when bussing though a folder a panned child track would be lower by the pan LAW, which is unexpected behaviour.

As was mentioned above it's literally just levels in a DAW when panning, perhaps in consoles and the analogue world there is some phase shifts as well as components drift per channel over time.


M

[GrizzlyAK]

Originally Posted by DrTT

Hello Schwa and Thonex,

thanks for your hints towards the discussion threads!

The main message from those threads seems to be that the pan law is applied nonetheless to "stereo tracks" and hence this distorts the amplification for the panned-to channel.

Unfortunately this makes it hard to emulate that behavior externally, so I am resorting to writing a simple JSFX-plugin that behaves in the naive way I like. It's a pity that then I cannot use the standard controls in the control panels, but I can live with that.

Nevertheless the non-linearity of the panned-from channel and the DC-offset should be fixed some time. But I have to admit that I have a very academic and measurement-oriented approach to music production, hence this might not be time-critical or even relevant to others.

Best regards,
DrTT

+10 to what DrTT said. I'm here trying to wrap my head around how Reaper's panning works NOW, in 2024. I, too, am very "academic and measurement oriented", and would love to see plots of at least the calculated voltage behavior for the various pan laws used in Reaper published somewhere, perhaps in the manual (maybe they have been and I've yet to find them). There is over a decade of measurements in these forums, few of which are likely relevant today. I know once it leaves the speakers all bets are off, but at least we can understand exactly what is happening up to that point. Right now, it's a black box.

Originally Posted by norbury brook

exactly, the issue with Reaper was though that when bussing though a folder a panned child track would be lower by the pan LAW, which is unexpected behaviour.

As was mentioned above it's literally just levels in a DAW when panning, perhaps in consoles and the analogue world there is some phase shifts as well as components drift per channel over time.

M

Yeah, I'm seeing this too, but this folder phenomenon is not JUST panning, though. In my case, I have a 5.1 audio item on a track that I run through JS Channel Mixer (great plugin by the way) to get 2 channels, which is inside a folder. The folder levels are about 3dB (estimated because this is similar to white noise) than the 5.1 track itself. So WHAT is happening here? I can't find anything in the manual about signal flow in folders, but clearly something is happening behind the scenes. In Audition, I found that converting a 5.1 (6 ch) file to Stereo reduced the overall volume by 8.15dB (empirically determined using Audition's AWESOME Amplitude Statistics panel). I suspect this is something related to 5.1 tracks being "converted" into stereo tracks, although they are all the same kind of track (I REALLY wish Reaper had true mono tracks :-/ ). It would be nice to have a complete definition of the signal flow and any impact folders have on it. Having to seemingly randomly boost folder tracks to make up lost levels is a little unnerving. What's really going on here. Anybody know?

EDIT: found https://forum.cockos.com/showpost.ph...32&postcount=7 and https://forum.cockos.com/showpost.ph...36&postcount=8 from Schaw.