Noise VST Plugin / EQ?

Spikeh

Spikeh

New member
This is entirely my fault, and I've learnt from my mistake, but I've managed to record a 7 hour voiceover (yes, 7 hours... plus about 10 hours editing :() with the gain on my preamp set too high :( This was due to me not spending enough time setting up - the SM7B mic was too far away from the vocalist and I didn't register it :(


So, the full 7 hours of recording is full of noise. It's not TERRIBLE, but believe me, I can't put it on a CD like this! I've been messing about with Waves XNoise and XHum, and I've got some decent results. Also tried Nuendo's built in Noise Filter, which is ok, but not quite adequate.

Are there any other techniques I can use to get rid of the noise? EQ maybe? As I've not needed to work with noise before, I've no experience with it and I'm finding it hard to remove the noise without losing some of the vocal quality / the sound distorting / modulating. Any hints would be massively appreciated.
 
Baaahahahahahaaa!!!

That totally sucks, man. Sorry to laugh at you (it's all in good spirits), but some lessons are just learned the hard way, I guess.

The Waves plugs are about as good as it gets for dealing with an impossible-to-completely-deal-with problem like you have on your hands. Don't you have SoundForge, too? There's a pretty good noise removal plug included with that, I think. Might wanna give it a shot.

Post a small chunk of it and I'll take a look at it for you later if I have time and tell you how I would go about removing your particular noise while trying not to butcher the actual sound.
 
It's at the studio (I'm at my DAY job at the moment, saving up for that custom built studio I was on about ;P), so I'll get it this weekend and put it online for you, ta :)

Tries the Sony Soundforge noise filter too... it's too harsh, regardless of the attack and release I put on it. The Waves one and Steinberg ones seem to be the best for it at the moment.

Would be helpful to know how I can go about analysing where the noise is too? i.e. what frequency it occurrs at? There's a "Learn Noise Profile" option on the Waves XNoise plugin, but I'm not sure how best to use it. Suppose I should read the manual, heh ;P
 
Without hearing it I can't be completely sure, but most "noise" is all over the spectrum. If it's the common 60 hz electrical noise - it will start at almost exactly 60 in my experience and be present at pretty much every imaginable harmonic multiple of 60 hertz....which SUCKS. Like I said - it's friggin impossible to completely deal with it (other than getting your stuff straightened out from the beginning, lol).

Harmonic multiples of 60 hz would be the octaves (like 120, 240, 480, 960 etc...) Most of the noise will be the octaves of 60. However, it will also rear it's hideous head in other harmonics - which makes completely dealing with it ....impossible.

If you want to find the worst spots of it - pull up your track EQ in Nuendo - dial the Q on one of the mids to 12 and boost it to 24. Now slide it slowly all the way across the spectrum until you find the most horrible, god-awful, make-your-ears-bleed sound you possibly can. Then start widening the Q and listen very carefully as you do this - if widening the Q makes it sound worse, then keep widening it until you actually hear good sounds start coming out (as in non-noise, actual vocal sounds) then narrow the Q back down until it's strictly horrible noise. Then drop it to -24, and stop the sound for a second to give your ears a chance to recover to normal. Now bring it up very slowly until you hear that noise you just cut out coming back into the mix - at that point, back it back off to where it's just out of your audible range.

That's the best you can do...and you can do that over and over again to your heart's content by inserting more EQs but you'll never get it completely without somewhat butchering the sound, sorry to tell you. Good luck with this, you're gonna need it!

LMAO@7 hours of noisy recording going unnoticed! (sorry...can't help it - that shit is funny.)
 
Before trying de-noise plug-ins and EQ; what about a good old fasioned gate?
That should get rid of the noise between the vocals.

An interesting technique I have used before is to copy the track many times and use a multiband compressor and EQ to block out all but a small frequency range in each copy.
This gives you a fighting chance of getting rid of noise without completly destroying the vocal as you can now treat each frequency band that you have isolated seperatly.

Concentrate on the 400Hz-600Hz area, as this is where a lot of human speech lives.

Saying all this, ultimatly if you want a clean track, you have no option other than to re-record.
 
Codmate made some excellent suggestions - and I also just remembered one more thing I've done before to deal with nasty high-end noise.

I don't know if you have Wavelab, but Wavelab comes with a pretty cool plugin called the Spectralizer that creates artificial harmonics - I am sure there are others like it, but it's the only one of it's kind that has ever come with any software that I've bought. I guess it's similar to an exciter perhaps...but I've never been able to use those like I can Spectralizer.

Anyway, I doubt this is the kind of noise you have, but if you have high end hissy noise on a track, you can cut the highs completely out (where all the noise is) and then reconstruct an artificial, noiseless, high end for the track using spectralizer. It's pretty badass because it has actually worked for me a few times on tracks that were too close to deadline for retracking to be an option.
 
Adobe Audition has the single best noise removal effect I've ever used, and I've tried Waves'. For vocals you can get rid of a lot of the noise without effecting vocal quality. It's an entirely different thing when trying to remove noise on acoustic guitar tracks or other full-range instruments.

I imagine the 7hr WAV file is way too big to send to anyone over the internet...I'd give Adobe Audition a try if you can, and check out their noise reduction effect.
 
Thanks for all your suggestions :)

I'll give the EQ methods a go...

Already got a gate on it, works fine for when there are no vocal, but the noise is so severe that when she breathes it sounds horrible.

LMAO@7 hours of noisy recording going unnoticed! (sorry...can't help it - that shit is funny.)
Click to expand...

Yeah, hillarious, especially for me ;P Nah... seriously, we recorded 30 mins initially, with the mic in completely the wrong position, no pop shield (well it was on, but she was speaking around it!), and the compressor was set up completely wrong.

Then I adjusted it, started recording again and the sound was a lot better, but I didn't notice the noise. I noticed it about another hour in, and by that point I couldn't change the settings. The client is in a massive rush to get it out - been mixing every night this week! There's no way we can re-record!

It's all happened because it's my first commercial client, and I was flapping as he was in a rush, plus the session before I had to send him home as my PC was blue screening (sorted out now ;P). I won't do it again, I guarantee that!
 
Spikeh said:
This is entirely my fault, and I've learnt from my mistake, but I've managed to record a 7 hour voiceover (yes, 7 hours... plus about 10 hours editing :() with the gain on my preamp set too high :( This was due to me not spending enough time setting up - the SM7B mic was too far away from the vocalist and I didn't register it :(


So, the full 7 hours of recording is full of noise. It's not TERRIBLE, but believe me, I can't put it on a CD like this! I've been messing about with Waves XNoise and XHum, and I've got some decent results. Also tried Nuendo's built in Noise Filter, which is ok, but not quite adequate.

Are there any other techniques I can use to get rid of the noise? EQ maybe? As I've not needed to work with noise before, I've no experience with it and I'm finding it hard to remove the noise without losing some of the vocal quality / the sound distorting / modulating. Any hints would be massively appreciated.
Click to expand...

Don't use X-Hum, use X-Noise.
You want to select a very small part of the noise with no vocals, just the pure noise. Then push the "learn" button while the noise selection is looping.
Then un-push the "learn" button, and then set the threshold and reduction while playing the vocal track.

Eck
 
ecktronic said:
Don't use X-Hum, use X-Noise.
You want to select a very small part of the noise with no vocals, just the pure noise. Then push the "learn" button while the noise selection is looping.
Then un-push the "learn" button, and then set the threshold and reduction while playing the vocal track.

Eck
Click to expand...

Aye, that's what I figured... and exactly what I've done with it! Comes out a bit modulated and lose a bit of highs. It's the fine tuning after the "learn" part I need help with :)
 
Spikeh said:
Aye, that's what I figured... and exactly what I've done with it! Comes out a bit modulated and lose a bit of highs. It's the fine tuning after the "learn" part I need help with :)
Click to expand...

You want to use as little as possible noise reduction, its a compromise between sound quality and noise.
You definitely don't want to hear much artifacts like robotized sounding vocals.

Start with the noise reduction at full, and then move the threshold so you can blatantly hear the plug in at work. Then bring down the noise reduction until it sounds natural and with a bit less noise.
This is your starting point, now you need to play with the NR and Threshold a good bit till you find a happy medium between noise and quality.

Eck
 
ecktronic said:
Yeah thats a good one MH, but for 7 hours of audio I think its out of the question! :)

Eck
Click to expand...

Aye... 7 hours is a little too long to automate I think... and I can't charge the client for it as it's MY mistake! Idiot...

Ah well, first bit of voiceover work, you live and learn :) At least I'm safe in the knowledge that my mix already sounds better than any of his previous mixes :)

I'll just continue faffing about with the noise plugin / EQ till I get it right. Need to wait for him to come back to me with any issues first.
 
Record voice overs in a vocal booth!!!!!!!!!!!!! :)
If you don;t have a vocal booth then hang up duvets in the corner of a room and sing/speak into the corner.
And always remember a pop shield!!!!!!!!!!!!!!
They don't take away from the quality of the sound really.
I tend to stay away from the foam things you put over the mic, I prefer a pair of tights stretched over a shaped-into-a-circle metal coathanger.

Eck
 
If only I had a vocal booth eh? I already use the blanket technique, hung over mic stands :)

I use a pop shield and the foam windshield that comes with the SM7... sounds fine with them both on. I just had her sat too far away from the mic ;P

I desperately need a vocal booth, see my other thread about it... I kept picking up car exhausts, doors banging in the building, bands playing down the hall etc... utter crapness.
 
Haha, it's 7 hours of recording... bearing in mind 1h of it is a re-recording of some sections, and taking in to account the mistakes that are made + discussions about script changes etc (I don't stop the recording, no point at that level), it adds up pretty fast! The end result is about 5h of material. It's about a 3GB file at 44KHz (another mistake I made - should have recorded it at at least 48KHz and dithered it down).

They're training CDs on finance, taxation, trusts, inheritance, wills, legalities etc... I'm learning a lot about the law at the same time, but it gets boring, and it's hard work... sitting there for 7 hours recording, marking mistakes on a script, then spending 10 or so hours mixing it in, removing mistakes (removing hiss ;P) etc. At least it's a decent amount of money ;P

All the editing is done now, just waiting for him to come back with any mistakes I've missed, errors, blips, timing issues etc... I offered to listen back to it in full myself and get rid of all the mistakes (I merely went straight to the mistake points I'd written down and edited them out, so I may have missed some during the recording), but he's happy to do that himself. Either way is good for me ;P

Not mastered it yet... I'll do that once I've got rid of all the mistakes etc and he's happy with it. Also have to do some copying + pasting for the final module (all the key points condensed).

Any suggestions on the mastering chain? This is what I'm thinking:

Waves XNoise -> Waves C1 Gate -> Waves Parametric EQ (10 band) -> Waves C4 Compressor -> L2 UltraMaximiser (or one of the limiters on there at least).

Also... what level should I normlise the track to? I would imaging if I normalised it to 0Db (or -0.03Db like I do normally) it would be too loud for general home / car use?
 
Spikeh said:
I desperately need a vocal booth, see my other thread about it... I kept picking up car exhausts, doors banging in the building, bands playing down the hall etc... utter crapness.
Click to expand...
Sounds like a right nightmare!

Eck
 
Spikeh said:
Any suggestions on the mastering chain? This is what I'm thinking:

Waves XNoise -> Waves C1 Gate -> Waves Parametric EQ (10 band) -> Waves C4 Compressor -> L2 UltraMaximiser (or one of the limiters on there at least).

Also... what level should I normlise the track to? I would imaging if I normalised it to 0Db (or -0.03Db like I do normally) it would be too loud for general home / car use?
Click to expand...

Funny you mention it. I just did a VO here for an on-boarding presentation. About 6 hours editing screw-ups and chopping the audio up into separate files.

My experience with an SM7 is that you have to get pretty close to the mic or it sounds thin, so I'm thinking that you're gonna need a bump in the botton end to compensate. Since there is so much background noise I think that I would shy away from compression and rely more on the limiter rather than a combination of both. Using both is going to tend to bring up the background noise further.

Without hearing it, I would suggest XNoise (to get rid of broadband noise, ain't gonna help with the band next door), possibly the gate (depending on how much this is gonna "chatter" given the type of background noise), EQ, and a limiter. I would shoot for a maximum RMS level of -10 dBFS on the voice.

Another thing that I would suggest is background music to cover-up the noise, and to make 5 hours of someone talking a bit more interesting. I'm sure that there are a few Metal artists here that can donate something appropriate to this material. :)
 
You must log in or register to reply here.
Back
Top