Compression when recording vocals?

[floydianite]

Can anyone recommend an inexpensive (< $150 US) compressor that would be good for vocal recording? I currently run vocals through an ART Levelar when recording. It's not bad, but I'd like to have more control over ratio and attack/release settings.

More importantly, do you think it's better to record vocals with or without an outboard compressor? Personally, I like to keep recording levels fairly hot (between 0 to -10 dbFS), so I need to keep transients and louder parts from clipping. Then I do most of the compression/limiting during mixing.

Thanks!

[james0tucson]

Two considerations: 1. It's common for singers to perform differently (maybe better) if they hear a wet sound as they sing. 2. There's an argument for recording dry and doing *all* fx as a post process. (This might also encourage you to fix the recording environment so that you don't need so much fx in the first place.)


Is there a $150 hardware compressor that's actually better than Kjaerhus? (Is there a $1000 one, for that matter?)

[where02190]

$150 compressor is only going to degrade your signal flow.

Stop recording so hot. 0dbfs is in the neighborhood of +28dbu, at or beyond the capabilities of most audio gear. Now multiply that by however many tracks you got and sum it, and you're overloading pretty massively. Record 24 bit, and target your ADDA's 0dbu reference as your nomial level. You'll have plenty of quality signal.

Work on getting a good vocal sound with the simple chain, mic-pre-AD.

[ahjteam]

http://www.kvraudio.com/forum/viewtopic.php?t=179949

[Landphil]

Worth mentioning, IMO, is the FMR Audio RNC
That is if you're looking at a possible hardware device.

It's about $50 bucks over your budget, but it's a great little piece. Can be very subtle or over the top, depending on your needs at the time. They also make a nice leveling amplifier and preamp as well. All are good bang for the bucks.

I use a UA 6176 for my main preamp and employ slight limiting while recording, but I try to keep limiting and/or compression at a minimum while recording vocals (most of the time). Just enough for good gain without clipping and squishing dynamics. You can always squish it more later, but you can't undo a squashed squish.
Mic technique is extremely important as well. With proper gain settings and knowing when to "eat" the mic during quieter passages and when to back off when you let loose and wail...

I go both ways. Some tracks I want as pure as I can get them, but some tracks benefit from pre-prosessing, depending on the nature of the song.

[floydianite]

Thanks to everyone for the feedback . I'm sure this has been discussed to death before, *ahem*, but I'd like to clarify some things where02190 said:

Stop recording so hot.

I should've pointed out that I was talking about peak rather than average level. I like the signal to peak in the range of -5 to -1 dbFS. Average rms is probably more like -12 dbFS.

0dbfs is in the neighborhood of +28dbu, at or beyond the capabilities of most audio gear. Now multiply that by however many tracks you got and sum it, and you're overloading pretty massively. Record 24 bit, and target your ADDA's 0dbu reference as your nomial level. You'll have plenty of quality signal.

First, I'm not sure what you mean by targeting my ADDA's 0 dbu as my nominal level. My setup is basically mxl2003 condenser -> m-audio dmp3 -> art levelar -> echo mia. Everything else is "in the box." Aren't dbu's an irrelevent reference in my case?

I record only one track at a time at 24 bit. Mixing in 32 bit float, I adjust (attenuate) the levels of each track individually. So I don't know how "overloading" would be a problem. It sounds to me like you're saying I should shoot for around -28 dbFS, which seems too weak. Keep in mind I'm far from working with "pro" gear or recording in a pristine studio environment.. very far in fact . I think noise-floor would be too much of an issue to record at such a low level.

Recording hot means I don't have to boost gain which I feel can degrade signal quality. I know I'm probably wrong here, but I've always felt more comfortable starting with a strong signal and lowering it in the mix. If I need to, I can use compression to boost the overall loudness.

[ahjteam]

Recording hot means I don't have to boost gain which I feel can degrade signal quality. I know I'm probably wrong here, but I've always felt more comfortable starting with a strong signal and lowering it in the mix. If I need to, I can use compression to boost the overall loudness.

You should record so that the signal-to-noise ratio is the best possible, on most analogue consoles marked as the zero-point. if the signal goes too hot, just drop the gain.

[where02190]

I should've pointed out that I was talking about peak rather than average level. I like the signal to peak in the range of -5 to -1 dbFS. Average rms is probably more like -12 dbFS.

Which is way too hot. that means you've recorded signals that are hitting +24dbu.
Don't look at your peaks, look at your nominal. Target your converters 0dbu reference, if you can't find it in the specs then target -18dbfs.

First, I'm not sure what you mean by targeting my ADDA's 0 dbu as my nominal level. My setup is basically mxl2003 condenser -> m-audio dmp3 -> art levelar -> echo mia. Everything else is "in the box." Aren't dbu's an irrelevent reference in my case?

Unfortunately they don't list the 0ebu reference on the specs online, but you could call or e-mail the manufacturer to get it.

Set your input level so the average (nominal) input level is in dbfs the 0dbu reference of your converter. If you can't get that spec, use -18dbfs. Typcially the 0dbu reference is between -20 and -12dbfs. This yields tracks that are all nomially hitting 0dbu, yielding a quality signal with tons of headroom. Unless they dynamics of the player are total shit, you'll never, ever come close to overshooting 0dbfs.

dbu is a reference, same as dbfs, it is not irrelevant. In fact it is a major reference level you should be aware of.

I record only one track at a time at 24 bit. Mixing in 32 bit float, I adjust (attenuate) the levels of each track individually. So I don't know how "overloading" would be a problem. It sounds to me like you're saying I should shoot for around -28 dbFS, which seems too weak. Keep in mind I'm far from working with "pro" gear or recording in a pristine studio environment.. very far in fact . I think noise-floor would be too much of an issue to record at such a low level.

It's a common mistake to think that just because you're in the digital domain you can't overload, However there is a very audible difference in mixes that are properly gain staged in the box vs those that are not. While on paper they should sound identical, in reality there is a difference.

Recording hot means I don't have to boost gain which I feel can degrade signal quality. I know I'm probably wrong here, but I've always felt more comfortable starting with a strong signal and lowering it in the mix. If I need to, I can use compression to boost the overall loudness.

You're already boosting the gain past where it should be. If you need to hear the tracks louder, turn up your monitors. All your tracks hitting 0dbfs (+18dbu on your converters) times 20-30 tracks summed, your faders end up at the bottom in your DAW or you are redlining your mix.

Understanding proper gain staging and applying it to your recordings from start to finish will make a very noticable difference in the clarity of your mixes.

[floydianite]

Thanks where! I think I understand a little better now.

From the echo mia spec page:

Nominal Input Level: +4dBu or -10dBV (software configurable)
Maximum Input level: +18dBu or +4dBV (software configurable)

So, with balanced cables, if +18dBu = 0dbfs (max level), then nominal input (+4dBu) would be -14dbfs. And 0dBu would be -18dbfs. I think. Which is exactly what you suggested.

Bottom line: If you need compression/limiting to keep the signal from clipping, you're recording too hot.

[where02190]

Bottom line: If you need compression/limiting to keep the signal from clipping, you're recording too hot.

Not always, but you've got the right idea. Some "vocalists" just have shitty dynamic technique, and other than riding the mic pre, there's no way to keep them from the red. However you can, if you are a trained vocalist yourself, teach, or attempt to teach them, some better technique, and maybe educate them toward improving their talent.

Also, often songs have a different dynamic feel between sections. In cases like this, if I possibly can (ie if the vocalist is willing) I track them separately, and adjust the pre accordingly, doing the loud and the quiet passages on separate tracks.

Kudos to you, you've comprehended some basic but vitally important methodology on recording in the digital domain. Best of luck.

[Muzik 4 Machines]

dbx tabletop 166, 119$ (mc6 i think is the model name)