Cubase 9.5: how good is vari-audio?

Hi all

As a long-time Cubase 7.5 user, I have only used vari-audio to re-tune guide vocals. but always do the final article in Melodyne - which I hate doing as it means having to record the audio into Melodyne plugin, and do everything through it’s slightly annoying interface.

So is the Vari-Audio quality in Cubase 9.5 (Pro) better than 7.5, and good enough to finally ditch Melodyne?


Melo is «better» in ways of more functionality and not destorting Sound. So no, you cant ditch melo.

Thanks. It’s just for tuning male vocals, nudging things up and down to the correct pitch. Not for large moves for creative reasons (like pushing things up or down a third or more)…Vari Audio in 9.5 still not good enough? And is it actually better quality than 7.5?

I don’t think 9.5 it different in its algo from 7.5. The handling has got better/easier, more than one event can be displayed in the same sample editor (maybe that was already in 7.5, not sure).

For small moves VA is easy and does a great, quick, well integrated job. It’s much like the baby version of Melodyne with the basic toolset.

Sooner or later Cubase will get ARA implementation which bypasses the transfer process and makes using Melodyne just as troublefree as VA already is.

Hi all

I use VA ALL the time particularly for crap singers who want you to make them sound good. I did use Melodyne just before Cubase got VA, but the convenience factor meant that I stopped using Melodyne completely. I find VA to be great for most jobs and if it doesn’t sound right, redo the vocal, I think Melodyne is probably slightly better because after all that is their main focus, but VA is still pretty damn good.

Best Regards, Dave

Melodyne simply goes far beyond pure tuning, nothing like it. The current version is one of the weirdest sound design tools ever heard, psychedelic stuff :wink:

VA is really good at simple tuning as well as strange tasks. It’s good to get to know all its possibilites to unleash its potential. So, different workflow, both great. Give us ARA and it will be editors’ heaven!

Melodyne is functionally a Swiss Army knife of course, but for the task the OP asked about (small pitch shifts for bad singers), is VariAudio just as good?

I use VA for that purpose often (male vox most often), and it is quite transparent.

I’ve read that Melodyne actually messes with the high end when processing … anyone here with that experience?

Tip: To get the best out of VA, first cut round the sibilants and always leave their segments’ pitches untouched. Once this is done, VA is really pretty clean.

I’m with “the parrot” - let Cubase detect segments and then manually delete anything that doesn’t represent the core pitch of the note (breaths, plosives, swoops, vibratos). Then correct the pitches from there. I can often get good results.

However, I’ve never used Melodyne so maybe I don’t know what I don’t know. :wink:

Yes, definately better to take sibilants out and leave them untouched. Unfortunately this can be a piece of work - it is in Melodyne as well!

There should be a better detection plus more automatic handling of ‘ssssssshhhssssssssssssss’. Something like ‘select sibilants only’, that’d be the most luxurious deesser of all times :wink:

Revoice Pro does a great job cutting out non-pitched elements.

But in VA: for a quick and easy way to make a bad singer sound good bit not robotic: select all, use your quantize slider to bring the average pitch as close to perfect as you want, then select and undo the parts that might sound off.

I find the things that require undo might be a few swoop in and out transitions between widely spaced notes, or restoring vibrato on some selected notes for realism (and then manually sliding them back to an acceptable pitch).

I’d never do it that way for a good singer, but the folks I record absolutely love the results, and I’ve found that undoing the effects of the quantize slider on some select notes is a million times quicker than tuning reach errant note.

Just curious - how are you performing this ‘undo’ operation please…? Are you literally just sliding back the quantise value previously set, on those notes that sound off, in one go…? Or is there some magic ‘undo’ button I am missing, that you can keep using on sections of the song, until you’re complete…?


After “pre-treating” the vocals like that, I listen through, stopping where it sounds wrong, figure out which segments are involved and address it from there. Most often it’s a Cher/T-pain kind of thing because there was too much of a slide into or out of a sung note, and the amount I’d straightened out the slide/translation was too much. So, for example in that case, I’d sekect the segment that had the slide, and then straighten slider to the left.

If when I first start checking out the vocals I notice a big slide from one note to another, I’ll make sure that the slide itself gets its own segment as much as possible. That can always be done after the fact too.

Another thing that almost always needs fixed with a challenged singer is that the autoquantize occasionally brings the note to the wrong pitch because the sung note was more than 50 cents off. Easily heard, and easily fixed, just slide it back to the intended note.

But I do all this one wrong-sounding note at a time, not in “one go” as you wrote. Even though I wouldn’t say it’s quick to do this, I really only need to “undo”/fix a relatively small percentage of the sung notes. It’s quicker than not “pretreating” the vocals and therefore having to adjust a lot more notes (with the singers I record).

Hope this isn’t too confusing - it’s late here, and I’m tired … sorry!

No, no… many thanks for your time alexis.! That’s why we come to these boards, to share and to learn - and those are all great tips from you, for anyone reading, when using VA.

I have to say, I already use quite similar methods in my workflow - I guess it was just that phrase you used ‘undo the parts’ that prompted my curiosity; that there may be an even quicker way to ‘backtrack’ on your edits I’d somehow missed… :wink:

Anyway, all good stuff - and always hoping that SB add more enhancements to VA as each new version update comes along; but that may be a pipedream now, as they come to the conclusion that simply adding ARA support has many win-win aspects, for both SB and users alike.


I’ve found the the following methods help to get a good result with Variaudio:

  1. Mostly avoid using the ‘Straighten Pitch’ slider and rely almost always on ‘Pitch Quantize’.

  2. Trim a lot of the Variaudio segments so that they are only covering the middle area of the note/syllable. This leaves the start and end of the syllable untouched and natural sounding so you can tune quite hard without sounding TOO artificial. It also avoids the ‘T-Pain’ effect on transitions between different notes.

  3. On a long sustained note, it can be helpful to split the segment into a few different pieces so you can quantise them individually/differently, enabling you to control the tuning without having to resort to the ‘Straighten Pitch’ slider.

I love Variaudio and use it a lot.

I have the all singing/dancing version of Melodyne, but 90% of the time I can’t be bothered to fire it up. Vari-audio is fine, and so convenient.

I would like an ARA2 implementation, though.

Absolutely can’t go wrong doing that!

I find that I can save a lot of time though, tuning quite hard for all the notes, and then just splitting off and “untuning” the relatively few times the TPain stuff shows up. I record some very amateur singers, and even with them I’ve found it takes less time for me to do it that way than a priori splitting out and blocking all the transitions from the VA process.

I’ve got both and other than the polyphonic element of melodyne which is admittedly really cool, if you’re just using it to tune a monophonic signal ie: a voice, I actually prefer using Variaudio. If you put some time into it you can get really clean results and it’s already there, build into your audiotracks in Cubase. I think it’s great. I use it on every session.

Another thing thats invaluable is being able to add a chord track and putting the key of the song in the chord track. Then when you open VA, you can choose to view the note colors by chord. It will then display in Red the notes that are off. This makes for a quick tuning workflow.

Another thing thats invaluable is being able to add a chord track and putting the key of the song in the chord track. Then when you open VA, you can choose to view the note colors by chord. It will then display in Red the notes that are off. This makes for a quick tuning workflow.

I’ll do this on every project. mapping the song structure using cycle markers, and making sure I have a complete chord track. I often do this even before recording.

Then decide on a tempo, record a really short basic drum track with the right feel(drummer plays along with songwriter on piano/guitar what have you).

Record sketch version of vocal and guitar/piano, so everyone knows the lay of the land.

Then take it from there. Works for me every time(when working on new pop songs)