Reaper now has built-in oversampling

[Craig Anderton]

1 hour ago, LittleStudios said:

@Craig Anderton, I'm not sure what you're doing, but so far I'm able to reliably able recreate this "bug" by simply enabling "upsampling" on playback and activating it with the "2X" button.

Okay - I'm able to reproduce it! Originally, I activated and de-activated the 2X button while the file was playing, because I thought that would be a harder test. However, that did not cause any problems. What DID cause the problem is if I started playback from the stopped position while 2X was enabled.

Try this workaround:

1. Start playback with 2X disabled.

2. Enable 2X during playback. 

3. Without stopping, hit Ctrl+Home (go to start). Playback starts from the beginning, with 2X oversampling. Or if you want to start from the middle of a song, start with 2X off, enable it, then place the cursor where you want playback to begin, without stopping in between.

However, I don't know if that means it's really oversampling or just that the 2X button is lit. But it does seem to be doing its thing. 

So yes, this appears to be a bug, and it looks like it's sidechain-specific. I tried using a patch point instead of a send to drive the sidechain to see if that would make a difference, and I also tried sending to a bus and then using the bus output to feed the sidechain, but neither of them made any difference.

I know workarounds aren't the most elegant option ("why don't they just fix the darn thing?"), and you shouldn't have to click a button in the middle of playback to initiate a function, but the above does seem to work reliably. Besides, when reporting a bug, sometimes reporting a workaround gives a hint as to what a potential fix might be.

Thanks for your persistence, so I could figure out what was going on.

 

 

[David Baay]

On 4/29/2022 at 10:26 PM, Craig Anderton said:

What DID cause the problem is if I started playback from the stopped position while 2X was enabled.

Which would of course be the usual use case. Glad you got it figured. I just came here t post a link to the bug demo project I sent to the Bakers last year, but i guess it's not necessary now.

On 4/29/2022 at 10:26 PM, Craig Anderton said:

So yes, this appears to be a bug, and it looks like it's sidechain-specific.

Oh, wait, I guess it is. Here's a link to a project that shows the issue without sidechaining. See the Notes for a complete description. There were some other issues that are no longer reproducible in 22.02, but the delay compensation issue persists.

https://1drv.ms/u/s!AvHuw7srYo1slQ4ZYAPxRQOH5Sea

EDIT:  I just brought this file into my laptop to verify the download and realized this .zip file has an old version of the bug demo project that does not have notes. Suffice it to say that if you zoom in, you will see that the clip in Track 1 had to be slid 140 samples late to make it null with the copy on Track 2 that is going through the upsampled Channel Tools. Note, also, that per-plugin upsampling enables are  stored in the registry, not with the project, so Upsample on Playback/Render for Channel Tools has to be enabled on your system.

Edited May 1, 2022 by David Baay

[PopStarWannabe]

On 4/29/2022 at 8:19 PM, Craig Anderton said:

Inserted the TS-64 Transient Shaper on one track

Just a bit off topic, but does TS-64 (and all those old plugs) still work? I have a lot of the old plugs that came with Sonar.

Can you please tell me what version the .dll file of your TS-64 is? Mine reads "File version    17.0.2.306"

My last purchase was Sonar X2 Producer. I want to know if the plug has been updated after X2.

[scook]

The last version of TS-64 is 17.5.2.14.

All the plug-ins installed with 64bit SONAR X2 should work in CbB.

[bitflipper]

On 4/29/2022 at 1:18 PM, Esteban Villanova said:

Do you guys have a strategy for using up/oversampling? Use your ears and choose on a per-track bases? How to easily detect cramping in an eq, for example?

Yes, I do. My strategy is to not explicitly use it, not Cakewalk's feature nor any plugin's built-in oversampling option. An exception could occur someday, but it hasn't yet.

Granted, this strategy works for me because I rarely put myself into a situation where oversampling would be necessary. I realize that's not the case for everyone. Some people like hard clippers, for example. I don't. Some use a fast-attack compressor on bass for distortion. I don't. Some like extreme limiting and brick-shaped waveforms. I don't.

But that's just me. Everyone's entitled to their own preferences.

As for EQ cramping, I do not have any EQs that are subject to cramping, because it's not 2003 anymore.

[Craig Anderton]

On 5/1/2022 at 3:43 PM, bitflipper said:

Granted, this strategy works for me because I rarely put myself into a situation where oversampling would be necessary.

But the problem is, it's impossible to know whether or not oversampling will make a difference, until you compare original and oversampled versions. 

This instructional video (hey, it embedded...cool!) is only one minute, but it will show you exactly why I say that it's difficult to predict whether oversampling will have no effect at all, or a huge effect. In this case, it's not a variation on a sound, it's an entirely different sound - it's like a totally different synth preset.

 

[bitflipper]

I don't want to seem argumentative, Craig, as you've done a fine job of illustrating a case where oversampling offers a clearly audible difference. However, it's an edge case.

We have to be careful generalizing from edge cases, lest beginners read more into them than is justified. This, I think, is the hole many users step into: that because oversampling can make a difference in unusual circumstances, it must therefore be useful or even necessary much of the time. This fundamental misunderstanding, I believe, is the genesis of hyperbolic phrases such as "game changing".

In the few instances where I have encountered aliasing from internal processes, it's been a poorly-designed processor or synth. Find me a patch in Omnisphere that benefits from oversampling, and I will immediately concede your point without reservation.

[bitflipper]

Here's an example of a poorly-designed plugin. It's a screenshot I made for my 2018 review of DDMF's Plugin Doctor.

Two points:

1. You won't see this kind of result from a quality fx plugin. This is a particularly awful one.

2. Even though visually there is obvious aliasing, it's debatable whether reducing it via oversampling would make an audible difference, because the aliased frequencies are 70dB below the main signal. Even in this extreme example, you might not even notice it. In fact, this dreadful plugin is widely used and has fans who swear by it.

 

[LittleStudios]

This should make everyone happy.  Dan Worrall made a video on this exact topic. 

 

https://youtu.be/GjtEIYXrqa8

Edited May 4, 2022 by LittleStudios

[Craig Anderton]

9 hours ago, bitflipper said:

I don't want to seem argumentative, Craig, as you've done a fine job of illustrating a case where oversampling offers a clearly audible difference. However, it's an edge case.

It's definitely an edge case, no argument there! The main point I keep trying to make is listen, and if something sounds better, use what sounds better (which may be the version with aliasing).

It's easy enough to raise the project sample rate, render, go back to the lower sample rate, and compare. Most of the time there will be no audible difference. But then out of seemingly nowhere, you hear a difference. The problem I encounter is that there are a lot of variables, like what waveform the synth is using, and whether the real-time oversampling algorithm is as detailed as the offline one. I would like to think that I could simply identify particular synths as needing or not needing oversampling, but that hasn't been the case.

Same with amp sims. Sometimes it makes a difference on high-gain amps, most of the time it does not. But I still check anyway, just in case. 

[Craig Anderton]

13 minutes ago, LittleStudios said:

This should make everyone happy.  Dan Worrell made a video on this exact topic.

https://youtu.be/GjtEIYXrqa8

I've only seen a few of his videos, but I like them a lot. I think the biggest deal with Reaper in this case is being to do oversampling once for an entire effects chain.

That said...my primary instrument is guitar, and I want the lowest possible latency for real-time playing. My experience with real-time oversampling is that it causes a major hit to your CPU, and you have to increase latency to compensate. So, for the moment, I think I'll be sticking  to "render-listen-keep or toss." 

[PopStarWannabe]

On 5/3/2022 at 7:09 PM, bitflipper said:

this dreadful plugin is widely used and has fans who swear by it.

 

 

Can you please state what plugin you are talking about?

If you don't wanna state here publicly, please send me a PM

[bitflipper]

BBE Sonic Maximizer