VaiAudio inconcistency

There is one thing about flattening warped and transposed audio that I find incomprehensible. On page 369, in the manual, it says:

If the audio has been transposed, the dialog also contains the “Formant Correction” option. You can activate this, for example, if you want to avoid a “chipmunk voice” effect.

A bit further down, it says:

If you introduced VariAudio pitch modifications, the MPEX presets will not be available and the algorithm preset “Standard – Solo” is used automatically.

What is the rationale behind this. Why isn’t audio quality and formant preservation just as important regardless of which feature is responsible for the pitch modifications?

This, unfortunately, seriously negates the usefulness of the “VariAudio” feature as a serious correction tool!

VariAudio preserves the formants (to a certain extend) by design anyway. The formants option is therefore only available for other algos.

“to a certain extent”!?! What do you mean? Either it preserves the formants or it doesn’t. There is no halfway about it.

Furthermore “Standard – Solo” algorithm is an inferior algorithm, intended for realtime previews. The MPEX or élastique algorithms are recommended for final rendering. Don’t you agree that it is pretty daft that we’re not allowed to use algorithms intended for vocals with a feature that was specifically developed for vocals!?

‘To a certain extent’ is meant compared to dedicated tools like Melodyne where formant handling is more advanced and can be edited seperately.

Standard Solo is the only available algo for VA. No idea if this is good or bad, flattening it produces the same result that you hear in real time.

When timing is an issue it’s a good idea to do it before or after VA editing. Before: you can choose any other algo for stretching (Standard Solo may not be sufficient) > bounce the file > do your pitch edits. After: bounce your VA edits > use the algo that sounds best to you for stretching. Without the bounce between the algorithm will be changed to Standard Solo as soon as you move the first segment.

When VA doesn’t work for pitch shifting tasks I use Melodyne. It also does smoother stretching. But for many tasks I’m pretty ok with VariAudio to be honest. It’s right at hand in the sample editor, extremely convinient. Guess if we had ARA implemented, I’d always use Melodyne instead.

That’s just it. Realtime algorithms are optimized for realtime (preview) use and are always inferior to high quality algorithms.

Still it makes absolutely no sense in excluding algorithms that are specifically intended for vocal work from a feature that is specifically intended for vocal work!!! I don’t know how many ways I must say the same ting.

Why bother with the high quality algorithms at all? Why is it important to use formantpreserving high quality algorithms in other instances (ex. Audio/Pitch Shifting)? Answer: There is a difference in quality! The realtime algorithm is intended for previewing during the editing process, not for rendering!

It just as stupid as if Cubase had 32-bit float processing internally, but only allowed 16-bit (without dithering) recording! I assure you that no professional would look twice at Cubase, if that was the case.

This is a serious limitation, that must be rectified if VariAudio is to be taken seriously!


By the way. Forget about using the harmonizing feature. Unless you’re working for Alvin & The Chipmunks.

Well, don’t get me wrong. I don’t try to justify lo quality processing. Just told you about my experience of what Cubase does or does not. The solution is simple: if you like what VA does for you in real time, go for it. If you don’t you better use a third party product for the task.

Wouldn’t mind, if VA gets raised in quality in the future. Recently I worked in Cubase 5 on a project I got sent which was done in 5 and was a little shocked about the warp quality compared to 6/7. There were many real improvements happening over the years. VA is pretty good in my ears when the corrections are not too drastic, but there’s a lot of headroom for it to get better…

With that kind of reasoning one could wipe away anything Cubase does wrong. I much prefer that Steinberg fixes the problem and let us use VA to it’s full potential. They might if more users demanded they do, and less encouraged them not to bother.

And letting us use the tools that are already there wound definitely be a quick and efficient way. I find VA excellently executed. It much faster and easier to use than the Melodyne version (even if they were first). The fact that it is integrated into Cubase is a definite plus. If Steinberg would let us use the high quality algorithms, it would be as close to perfect as I believe is possible. Unless they could figure out a way to make it work with polyphonic audio, that is.

Cubase is the best DAW out there. That is the reason why I’ve used it for some fifteen years (before it was a DAW and just a midi-sequencer). It’s great pity that something like this would tarnish that image!

This was raised on the forum many years ago.
The official word from SB developers was that the the solo algorithm is the same quality as Elastique Pro.
That was clearly nonsense then, and it is still nonsense now.

edit: found the post, it’s here: http://www.steinberg.net/forum/viewtopic.php?f=91&t=3329&p=29600#p29600
and another: http://www.steinberg.net/forum/viewtopic.php?p=67106