robert_e_bone


Over time using V-Vocal, I found that for me, best results occur when not using the auto stuff, and just do things manually.

The other thing for me is that I get the best results doing very small clips.  It is certainly more a pain to do it that way, but I find it makes the processed section MUCH less goofy sounding.

With the above, it does an OK job of fixing slight vocal flat/sharp problems, and I sometimes either add or remove vibrato, but again small clips produce better results.

I hoper that helps - WAY kewl you have so many critters where you live - except for the saltwater crocs and snakes.

Best - Bob Bone

Yeah, as I've said, I never use auto. I don't think I've even touched those controls. I'm perfectly happy to do the painstaking manual work. It's the only way to get a natural result. I sit back and just LISTEN to the singing and only bring up v-vocal if I hear something that's not right. I will then only process it in a small clip on that section ONLY. If it's for a doubled backing layer, however, I get a bit faster and rougher with my edits as it doesn't matter.




The problem I have here is not how to use v-vocal. I've been using it confidently for years now. The issue is that it is glitching on me. I'm doing the most basic of basic manual edits and you can hear v-vocal cutting up the samples to make the pitch change. It has never done that to me before. This is a bug. It's not really a matter of audio quality of v-vocal, it's that it's not actually working as it's supposed to. Did you get a chance to listen to the clip I posted? You'll hear right away that it's not just v-vocal sounding 'auto-tuney'. It's actually glitching.


https://soundcloud.com/ma...-vocal-sample-clipping


Cheers Matt :)

I think I'm just going to cut my losses. It looks like it's a case of my singing having too many overtones in it in this section (or something like that) and v-vocal is just getting confused.

When singing this part, I only got about 2 good takes till my voice got a scratchy sound to it. I think some mucus was building up from the belting or something... That gives it a rough sound and takes away from the smoothness. The main LEAD part is smooth and v-vocal has no troubles with it. Behaves as I'd expect. The backing layers were recorded later on and my voice was on the way out a little. I'm pretty sure v-vocal just can't detect it clean enough because it's probably seeing it as not truly monophonic or something. A better quality algorithm would probably do the job fine.

Guess I'll just need to work on my singing to get it smoother next time ;)

So it doesn't look like it's a bug or anything that can be resolved too easily. V-vocal's algorithm just isn't up to this particular section I think.

Sounds to me like you're doing everything right, but sometimes there is just something about the audio that confuses V-V. 


Listening to your clip, the stuttering sounded like nothing I've heard out of V-V before. So I downloaded it to take a closer look. I thought at first it sounded like regularly-spaced dropouts, but that wasn't the case; rather, there is a beating oscillation superimposed on the audio.


I loaded your sample into SONAR and created a V-Vocal clip from the original, unedited clip. Sure enough, I got exactly the same results as you, confirming that the problem lies within the clip itself, as opposed to some other issue with SONAR or your system.


What you're hearing is a beat frequency whose pitch correlates to the amount of shift. It sounds as if there is a component that's not being shifted, causing it to beat against the shifted version.


Also noted, and extremely bizarre, is that if the shift is larger (2+ semitones), then the remaining audio is pitched up even though it hadn't been edited at all. V-V seems to get stuck in a weird mode after the edit point.


There is nothing obviously unusual about the original clip other than more 15KHz+ content than I usually see. On the theory that there might be some very high- or very low-frequency content that was perhaps beating against V-Vocal's 20Hz quantization frequency, I tried severely filtering the extremes. This had no effect. There is also a high-pitched buzz around 4.6KHz, but notching it out had no effect, either.

I loaded it into Melodyne Studio. It, too, played back with artifacts, although they were somewhat less noticeable than under V-Vocal. As soon as I shifted the pitch in Melodyne, the sound got thin, but there was no beat frequency. Melodyne does a better job of figuring out the component frequencies than V-Vocal, so perhaps this supports my theory about some component not being shifted, while one of its significant harmonics is, leaving the two to beat together and produce the oscillations we're hearing.


Bottom line is there's something amiss with that clip. I'd sure like to know what it is! 

Looking at the clip again, I now see that there is strong evidence of comb filtering, with notches in the spectrum at ~200Hz intervals. That's probably why the clip sounds kind of hollow. Whether it's relevant to the problem, I don't know. It does suggest that you need some acoustical absorption around your microphone, though.

Here's a spectrograph of your clip, a textbook example of comb filtering:
 

Here's the buzzy part, with a close-up of the clicks:

There is one more thing you can try Matt.

Copy the raw, clean, un-v-vocalled  clip to a blank project.

Do your V-vocal edits in there, bounce it down (if it's ok of course) then copy it back to your original project.

This has helped me overcome some of VV's quirkyness in the past.

Hmm, very interesting. Thanks for that, bit! I was hoping you would have a look at this post being the v-vocal guru.

Vox was recorded through my Rode K2 set to cardiod (though this is an infinitely variable polar selector switch, remember - a perfect cardiod is next to impossible as it'll probably have an omni or fig 8 element to it to some degree). I recorded it in this room (with the 'door' closed):





I wouldn't have expected any comb filtering to occur in this as it was pretty dead, but it did have a hard tile floor and of course couldn't do a lot for the lower frequencies. Not sure. My gut feeling it was just my voice not being super clear and putting out a second harmonic with my 'rough' voice kicking in a little due to strain of over singing a little.


I'm glad to hear you tried on melodyne and it didn't quite play happy either. I was very curious about that. I was considering looking for some free correctors to just fix this one issue and use v-vocal for everything else as it tends to work well enough. But I'm not sure if I'll bother now that I hear melodyne couldn't quite pull it off either.


The beating makes sense. I didn't think of that being the cause but you might be onto something there.


As for the audio getting stuck in that weird high mode even though it's not being shifted, I've had that happen to me once before.


http://www.soundclick.com/bands/default.cfm?bandID=624285

If you listen to the song Broken Angel at 3:12 you'll hear a second harmony in falsetto singing an octave above the lead. I never sung that part in falsetto. The was v-vocal being weird, but I actually really like it at the time so I left it! The song has since been re-recorded for the album though so that version will be gone.

Your second image appears to be broken so I'm not sure what you were referring to there..

Cheers,
Matt

Bristol_Jonesey


There is one more thing you can try Matt.

Copy the raw, clean, un-v-vocalled  clip to a blank project.

Do your V-vocal edits in there, bounce it down (if it's ok of course) then copy it back to your original project.

This has helped me overcome some of VV's quirkyness in the past.

Thanks man, I haven't actually tried this though I don't like my chances as it didn't work on bit's computer either.. I will give it a go just in case. You never know with some of these things! Considering that it is a harmony/double part so not critical, I might try some LPF and HPF and see if it behaves any differently with them bounced on the clip first.

I like your vocal booth, Matt. Necessity is the mother of invention.

It may sound very dead in there, but it's because the thin absorbers are sucking up mostly high frequencies, making it sound dead but actually having very little effect on lower frequencies. Below 500Hz you need to have LOTS of absorption to make a difference. I suspect your biggest problem may be the close proximity to the wall. Think about ways to maybe set up the booth further from the wall, or perhaps re-orienting things so that you're facing the wall when you sing (so that the wall's in the dead zone of your cardiod mic).

Matt, lol....I don't mean to laugh as I know this is no laughing matter. But that artifact totally cracked me up!

To compensate for my laughter, I'll gladly try to fix this track for you. :) Split the section you want fixed, hi-lite it and then export it as a broadcast wave in the bit and sample you recorded at without any dither.

Send it to dannydanzi at dannydanzi dot com and I'll take care of it per your instructions on where you want it altered. I've used VV billions of times for clients and have never run into the issue you're running into. So if you need me, just send it and I'll email you back with a broadcast wave of that section fixed without artifacts. :)

-Danny