Removing noise floor from spoken word via SM7

[christian_roth]

I know the SM7s need juice and I usually can minimize the noise floor with louder vocals. This time, I had a client who was doing a mellow spoken word thing and liked the sound of the SM7.

I have the SM7 going through an Avalon 737 cranked to almost full input and output. To combat the noise between spoken parts, I threw a gate on there. It helps but there still is some upper frequency noise on the parts which I tried to take out with some EQing.

I can still hear it but it's not as bad. Am I moron for choosing this mic for this style or just a moron for not finding the right tool to remove the noise? Help please.

[vvv]

You need a decent noise-reduction plug-in.

I have one in CEP; if you would like, PM me and I'll send you my email to send the file to me.

[Gregg Juke]

Assuming, as vvv did, that you are in the digital domain, you can also get _real labor intensive_ and edit (fade in and fade out) all of the heads and tails of each word (I mean tight, like on a near pixel/granular level), and then manually mute/insert silence between words, rather than use the noise gate. If it seems a little unnatural, you can play with just an infinitesimal amount of the tiniest little bit of reverb or room reflection, to give your word endings a little tail (you might even try that with the noise gate if things are a bit choppy).

There's always a way around; how bad do you want it? I just used the same process tonight (very successfully) on some very short (one sentence) voice promos that had a lot of room sound/background noise, and it took a couple of hours (well, maybe an hour and a half). A whole spoken word piece? Better make some coffee, and maybe a bag of popcorn...


GJ

[Snarl 12/8]

Record the mic cranked in the room and flip phase.

j/k

Edit: I just realized something. If you record the mic alone, or use a stretch where there's no voice you could find the shallowest, narrowest eq setting the makes the noise disappear and then apply that to the track.

Or, just send it to VVV. That's a hell of an offer.

[Gregg Juke]

Problem is the noise is often "multi-band;" you could try a notch filter or a bunch of them or maybe multi-band compression?? Phase-flipping and EQ (graphic or parametric) only does so much, and then you wind-up with not-quite-good-enough-of-a-fix, but terrible EQ settings for the actual program audio.

Tough spot to be in...

GJ

[MoreSpaceEcho]

all good suggestions, i'd probably try noise reduction first...if you're doing this sort of thing in the future, if possible put the singer in a quieter room...solve the problem in front of the mic rather than after it.

[kslight]

In addition to the above suggestions, you could try a De-esser maybe?

[hogfish]

The 737 is a nice channel strip, but it wouldnt be my choice with that particular mic. Sm7s definitely benefit from high output mic pres like the high gain Grace 101, Neve VR and 88R type preamps, AEA ribbon mic pres etc. Or maybe try a Cloudlifter in-line.The 737 definitely sounds great with Hi Output mics like condensers, but with just a touch under 60 dB of gain when open all the way, I dont think it has enough gain for the Sm7s low output...........

[vvv]

FWIW, on the less expensive (< US$400 used) side, I really like the SM7B with the ISA1, and the Summit 2BA-221 which both also work nicely with gain-thirsty ribbons.

[kayagum]

You can try using a Triton FET Head or a Cloudlifter.

I used the noise reduction in Adobe Audition for most of my sound design work including VO work- edfective and highly recommended.

Don't be afraid to try other dynamics...

[permanent hearing damage]

I just had this exact same issue recording a podcast! I used a combo of gating the top end via a dynamic multiband (shelf around maybe 5k) and then in parallel low passing the signal and squashing the hell out of it. I did give 10k a bump in the end to give it a little air back.