Sound quality

Steinberg Lounge
1

This morning, I bounced some tracks via Cubase 8 Pro amidst several crashes. This was an existing project I had done in 7.5.

Upon playback, they didn’t sound quite right to me. I got the WAVs I bounced with 7.5, and to my horror the difference was notable.

I checked my project in 8 to ensure all the tracks, levels and plug-ins etc were the same and present. Everything checked out. Indeed, it sounded very different unbounced.

I did a test with some dry tracks. What I found confirmed what I heard originally. The quality of the fidelity in the bounced via C8P was poor by comparison to C7.5.

To my ears, it just sounded awful bounced in 8. I checked the files with some tools to analyse the frequency curves and spectrum, and sure enough they are not close.

I’ve just done another test on a song and I’m astounded by the differences. Some of the differences are subtle, some more obvious, but this is totally unacceptable to me.

To me, this is a step in the wrong direction. It’s like they’ve gone one step forward and five steps back.

If this is the new Cubase, for me I can’t live and work with this. For now, I will use 7.5.

But if Steinberg don’t resolve these issues and improve the fidelity, I won’t be able to use 8. I’m disappointed.

When I stopped using Sonar a few years ago as my main DAW, I opted for Cubase against Pro Tools because I’d already used Cubase more than pro tools previously too. But this is so bad, and frustrating.

I want to use the features in 8 but whatever they have done to the code has adversely affected the product and fidelity.

I love using Cubase, but if this represents the future of Cubase, it’s not for me.

:frowning:

2

the first thing everyone will say is “does the 7.5 mixdown null out with the 8.0 mixdown”?

so you may as well get this arguement out of the way and render both to the same quality wav (eg both to 44.1k 24 bit) and see if both wavs cancel out completely …the “experts” seem to think this proves everything. i disagree but it would a good place to start.

(2nd thing they will say is that the fx plugins being used are different. eg reverb presets or other so double-check this.)

let us know how you go man. this is a potentially serious problem and dealbreaker for everyone im sure.

Cheers,
Blackout

3

Hi, that was the purpose of my dry test I did earlier today.

4

The wavs definitely didn’t cancel out

5

As far as I can tell from the quick test I just tried…dry tracks are nulling perfectly.
Tracks with insert effects are nowhere near to nulling.

Maybe pdc is not as accurate (or is more accurate?) than in C7.5

There was no apparent difference in quality.

EDIT: Tested some more… in a clean project I haven’t been able to get anything other than a complete null…with the exception of Slate VMR hiccuping on first hit after opening project (easily remedied by always playing the start before exporting)
Now I think about it some of my tracks in the first project I tested had audio warping (the ones I can hear most clearly after the null test) so I think changes to the elastic algorithm are causing the lack of null.

6

Definitely different on my systems - even more problematic with a VST (no other plug ins) as the source.

I checked the positioning in track by importing two separate instances of the same mono track from 7.5 and inverting the phase on one. They cancelled out fine.

I did the same with two from C8P and they cancelled out fine.

So, I know the samples are aligned ok…but when I take exactly the same from 7.5 and 8 and invert one, they don’t cancel out completely. In fact, there’s quite a difference… a fair bit of sound coming through.

I can definitely hear differences between 7.5 & 8. It concerns me that there’s a difference, and that it’s not a positive difference. I don’t hear the qualities and timbre in from C8P that I’m getting via 7.5 on individual tracks - especially on piano.

Visibly, the waveforms don’t dissimilar until you get in close.

Out of curiosity, I did the same test with Sonar. I found X3s end-result is closer to C8P than C7.5 in terms of how sound bleeds through. I didn’t expect that. :frowning:

Anyway, I bounced down again with a ready-for-mastering song with and without plug-ins. Definitely a much bigger difference with plug-ins on, but still negative differences with them off.

The low end seems lacking and the mid / mid-highs seems harsher and not as clear. Frequency analyser showed this two, lower response in the lows and a busier response between 1.5 & 5khz

But it’s more than that. There’s just something about the overall timbre of the sound that isn’t as good.

If I moved to 8 with it like this, it’s going to be a costly and time-consuming nightmare revisiting projects or working on current projects.

I played back some reference tracks via C8P a few moments ago and I heard similar problems there. For me, that sucks for referencing. It just didn’t seem to have the fidelity in C8P.

I’ve got two music machines (both PCs, Windows 7, similar specs), and they produce the same result.

I’ve been up since the early hours on this. There are features in 8 that I really want to use because of the benefits. But with this, it’s not worth the hassle it will cause me.

I’ve even switched out my audio interface and monitors today to rule that out.

I’m testing with a friend on his gear in about half an hour or so. His gear is different to mine, so I’ll see what that shows.

7

When 1+1 starts adding up to 3 … I’ll start getting worried!

Cubase Pro 8 sounds absolutely identical to 6/6.5/7/7.5!!!

I have never heard or been able to do anything other than null to complete silence files made in any version.

Oh for that matter Cubase also nulls to silence using identical files with Logic, Pro Tools, Studio One etc etc.
And yes I’ve done this myself - so I’m 100% certain of what I’m saying.

It’s that maths thing, kinda hard to escape isn’t it.

Now if your using different plugins in different versions or different plugins in different DAW’s that will introduce a variable, but for a given file, given fader setting, given pan (providing the pan law is identical) the result will be identical.

The core function of a DAW audio engine is to subtract and add numbers, it doesn’t arbitrarily throw a few in or chuck a few out because it’s in a bad mood, or feeling happy that day.

And I was guilty of thinking Cubase sounded better than Logic etc until someone made me sit down and do some proper testing and I proved to myself they all completely null to total silence given identical settings!!

8

Oops…added this as an update to my last post but as you posted in the meantime probably best added as a new post:

I tested some more… in a clean project I haven’t been able to get anything other than a complete null…with the exception of Slate VMR hiccuping on first transient after opening project (easily remedied by always playing the start before exporting)
Now I think about it some of my tracks in the first project I tested had audio warping (the ones I can hear most clearly after the null test) so I think changes to the elastic algorithm are causing the lack of null there.

Definitely different on my systems - even more problematic with a VST (no other plug ins) as the source.

VSTi pretty much don’t count in null tests…Round robin samples, free running lfos, etc etc.

9

Here’s an example of the differences. These subtle differences allow sound to bleed through on the null test.

However, there are more obvious ones I am getting.

And, as I said, these tests are dry.

Differences.jpg
Differences.jpg1195×649 76.8 KB

10

This is a piano I used in a null test.

If I have 2 instances of a bounced version in 7.5 and phase invert 1, they cancel out perfectly.

The same with 2 instances of a bounced version via C8P.

But one from 7.5 and one from 8, phase invert either and their are differences.

You can hear the sound bleeding through.

11

Link: File Dropper - Online file sharing

12

After bouncing out from the first version are you saving the project and re-opening in the other or are you creating a new project and importing the file?

13

Sorry, I should have said, been using both audio and VSTi, but I’ll stick to audio alone now for this.

14

Just been bouncing down with some plug-ins enabled on vox… I don’t like what’s coming out of C8P with my CLA-2A. I’ll upload after when I am back.

15

You have to start your null testing between an A and a B will very basic files under basic mixing scenarios.

They will null.

As you add things like piano’s and FX plugins all you’ll discover is (as pointed out) is those units add a variable.

Piano VI’s and the like add round robin samples, compressor plugins seem to be able to generate variable results due to the way their modelling is implemented in regard to compressor envelope curves and non linear amplifier characteristics.

When avoiding compressor plugins or any plugin designed to introduce non linear characteristics and of course round robin generating VI’s, your files will null to complete and utter silence.

16

Here’s the outcome of only a quick test we could do on a friend’s system with C8P & 7.5 - dry vox, all done in mono recorded with one mic

You can hear aspects of the vox coming through.

We tried Sonar X3 too… no plug-ins exported as a single track, the audio was null - but exported via a bus in X3 and it was not completely nullified.

Back to Cubase, we did a test with some recorded bass guitar, and that was null when tested.

But then the odd thing was in the 3rd test, guitar & vox (no VSTs or plugins), some sounds were coming through.

:confused:

17

:confused:

I don’t know what to make of it now. There’s definitely audio bleeding through on the null test in certain circumstances / audio. Frankly, somewhat ironically, it’s given me a headache :frowning:

Just had a listen to a few go-to plug-ins on some previously bounced audio, & bounced it down in 7.5 and 8. I much prefer how it sounds via 7.5 than 8.

But I’m tested out with it all. :frowning:

18

Here’s the outcome of only a quick test we could do on a friend’s system with C8P & 7.5 - dry vox, all done in mono recorded with one mic

Huh…are you saying this is now testing on a different install of C8 and C7.5 on a different computer?

I promise you that I am getting a null with any dry audio file here…& I mean an absolute null, not just inaudible but nothing registering on the meters at all.

I asked this already but let’s try again…Are you saving and loading the same project or are you importing the file into new projects in each version??

19

I just tried this and, in a 7.5 project (big mix, lots of plugins), I got a serious, noticeable difference in the overall tone and stereo spread of the mix! Much more low-end and wider stereo image in C8’s version of the mixdown.

I wasn’t trying a null test. C8 must be processing differently (or ignoring) some plugins. Very, very troublesome!

20

Forget worrying about made with “one mic” or using “bass guitar” … it’s just a digital stream.

Use the built in signal generator in Cubase set to -18dBFs 1KHz on a stereo track

Duplicate to make 8 (or 16 or 32 whatever just adjust for overloading the mix bus) of them in C7.5.

Mix those streams in C7.5 with some fades and pans … add some EQ using the built in EQ in Cubase.

Bounce down the resulting sonic mess :slight_smile:

Import the 7.5 project file into CP8 bounce the mix again.

Now null the two mixdown files and it will null to complete silence … at least it does on my system.
This just says to me on a basic fundamental level Cubase Pro 8 still manages 1+1 = 2

If you don’t get a complete null, yep I’d start worrying.