How to manage dynamic range

[MisterTim]

Sometimes learning everything I know off the internet doesn't work to well. Example:

I know how compressing works, I know how all the controls work and what they do, but I still can't use it effectively.

Last night I was recording a choir concert. This is both large group (with piano) and small a cappella group. I have no problem using compressors effectively (live) to control things like clapping or stomping. However, the dynamic range is so huge in my recording. The louds are normalized to where they should be, but the quiets are way too quiet. Compressing it in post-production obviously never sounds right whenever it's more than one microphone or pair.

As a musician, removing dynamic range doesn't sit well with me, but that seems to be the only way to make a CD sound decent. So is the solution to be a lot more severe on the compressor in the live production?

I'm sorry if this is a bit hard to understand, I'm really having trouble communicating and figuring out why I can never get my recordings to sound quite like I want them to.

Also, in case it matters, I'm using a Mackie tt24 and an Alesis ML-9600, so any tips you have in reference to utilising these pieces of equipment more effectively for compressing would be awesome.

 

[blalew]

If you have an Alesis Masterlink, you should get a reasonable recording. What are your settings? Seems if you have the ML that you can record the raw two-track to the hard disk, then play with some settings on it to get your recording sounding the way you want.

From Alesis' website:
What are the correct settings for compression and limiting?
That's a matter of artistic choice–there is no "correct" setting. But most professional mastering engineers use very subtle amounts of compression, sometimes taking only 1 or 2 dB off the peaks. To find out what you prefer, you can put the same songs into different playlists with different DSP settings, and listen to the resulting CDs for a few days through different systems. It's a great evaluation tool.

I almost always engage a compressor on vocals, but it engages when they get to about 80% of full volume so it's reasonably light. However, I use an Aphex Compellor to level/compress the recorded output. My point is that you'll probably need to add compression to your recording, despite adding compression for your live event.

 

[avare]

There is the tried and true old fashioned technique that became called "manual compression" when outboard compressors became popular with less professional users.

Ride the gain. When sections are quieter raise the fader, and when the performance is louder, lower the fader. If you are using a DAW program, you can program these adjustments into the mix.

Andre

 

[FMEng]

Andre is right. A slow hand on a fader is often the best way to handle choirs. The trick is not to try and keep the levels even. The trick is just to keep the softer parts from being too soft, and keeping the peaks from clipping. You still want the soft parts soft, and the loud bits loud. This results in a recording that still has the director's creative intent without it being hard to listen to in different environments.

I record a lot of choirs. Typically, I set levels so that the loudest passage in the whole concert still has 3-4 dB of headroom. I do not adjust the levels at all during the recording, unless necessary to prevent clipping. Then I adjust the levels modestly during editing with Adobe Audition. That way, anything I do to it is not destructive. If there are smaller ensembles I might crank up the level a few dB during recording, compared to the full choir.

If I know that I won't be editing the recording at all, then I use the slow hand on fader method. I try to let the softer passages dip down around 15 dB lower than the loud stuff. By anticipating the musical changes, you can adjust the fader so slowly that the gain change becomes almost impossible to notice.

Aggressive mastering of pop music and broadcasts have conditioned us to treat dynamic range as something wrong. In reality, music is supposed to have dynamic range, and the wonderful thing is digital recording can give it to us.

 

[blalew]

I'm sorry, my suggestion was based on the fact that you would be mixing for PA, then recording that.

So if your mics are just for recording, then record it w/ no compression & do it all in post on a computer. Just remember that compression means less dynamic range (duh) and with this noise floor is important - set your gain appropriately.

 

[MisterTim]

I'm sorry, my suggestion was based on the fact that you would be mixing for PA, then recording that.

So if your mics are just for recording, then record it w/ no compression & do it all in post on a computer. Just remember that compression means less dynamic range (duh) and with this noise floor is important - set your gain appropriately.

I'm mixing for both PA and recording simultaneously, but with different mics going different places for different groups. (e.g. large chorus doesn't need house reinforcement but piano does, small a cappella does does need reinforcement, etc.)

Also, I can't compress in post-production if I have more than one mic/mic pair, which is often the case with solos, accompaniment, etc.

 

[JoeGriffin]

As a musician, removing dynamic range doesn't sit well with me, but that seems to be the only way to make a CD sound decent. .

Well, you kinda have to. I don't have the exact numbers in front of me, but the human ear hears something like 140 dB of dynamic range, a CD can reproduce approximately 90dB...and LPs were even worse (about 45 dB of range). Capturing a live event onto these playback media pretty much requires that you squish it a bit.

On the other side of things, keep in mind what you're used to hearing from CDs. Especially with the Loudness Wars, you're probably hearing an extremely limited dynamic range on most commercial CDs--maybe less than 12 dB in some cases. So if your CD doesn't sound like commercial discs (even commercial classical discs), that might be why.

 

[FMEng]

I'm mixing for both PA and recording simultaneously, but with different mics going different places for different groups. (e.g. large chorus doesn't need house reinforcement but piano does, small a cappella does does need reinforcement, etc.)

Also, I can't compress in post-production if I have more than one mic/mic pair, which is often the case with solos, accompaniment, etc.

Doing both PA and recording can make it a lot harder. But, you can send a different mix to the recording using a pair of aux sends. If you don't want the main faders to affect the recording, set the aux sends to pre-fader. If you do want the PA mix to affect the recording, but with over all different balance, use post-fader aux sends. Use channel mute buttons when neither PA nor recording needs that channel, so that you don't have so many things to adjust on the fly.

I do tend to put some compression on solo mics using channel inserts. That is always helpful because most vocalists need it. That would be easy to do during the gig and would improve both PA and recording. Soloists probably contribute the most difficulty for you, as groups don't tend to be as dymanic.

Just throwing any old compressor or limiter on the stereo mix to the recorder isn't going to sound good. There are only a handful of units out there that could do that without making a mess with a lot of pumping and breathing. One of them is an Aphex Compellor which tends to be rather spendy, but for good reason.

 

[blalew]

Agreed on the aux send. I would go post fader, and it will take a good amount of experimentation to get it right, and assuming you can get it right, it probably won't be right for the next event. Recordings off the PA board are typically pretty difficult to get right when the room isn't a good representation of what's coming out of the master.

A similar situation that's probably equally difficult is when you're dealing with a rock band in a small/med room b/c the drums contribute so much acoustical energy on their own. Another tough one - you have an overwhelming wash of monitors into the house. Neither recording straight from the master will be very good, and an aux will take some tweaking during the event, which you don't really have the luxury of doing (it doesn't sound like).

If you're really interested in making this a decent recording, set your gains correctly and use a multi-track recorder (whether to computer or standalone like the Alesis HD24) to record everything - pre EQ/fader/dynamics/etc. But make a reference with your ML off a pair of auxes as a starting point.

We have a couple of tt24s and the lightpipe direct outs are great. You can use that to go into an HD24 (say) which will give you 24 channels @ 44.1/48, or for 32 channels into a computer you can use something like the Presonus Firestudio Lightpipe.

Sorry I'm finally understanding what you're trying to do There's no easy way, I'm afraid.

 

[blalew]

One additional thing while I beat this dead horse, on the tt24 you can link 2 auxes to make a stereo aux and then you can use aux mode (which puts the aux levels on the faders) & aux pan button to do set levels/pan on the recording aux, which is about as close as you'll get to easy. That plus the ML would get you something usable, possibly.

 

[MisterTim]

But, you can send a different mix to the recording using a pair of aux sends. If you don't want the main faders to affect the recording, set the aux sends to pre-fader. If you do want the PA mix to affect the recording, but with over all different balance, use post-fader aux sends. Use channel mute buttons when neither PA nor recording needs that channel, so that you don't have so many things to adjust on the fly.

Just throwing it out here, but I do completely understand how to run aux sends, pre and post fade, etc. I'm generally using a combination of pre and post fade, and for a show like this I'm almost always wearing my cans and touching up the prefade sends in the aux master. For a chorus concert the performance usually takes care of itself, just a few bumps on the solo mic in the house depending on the singer.

Soloists probably contribute the most difficulty for you, as groups don't tend to be as dymanic.

That's actually completely opposite of the problem I've been having. With the a cappella group I've been working with, the soloist generally stays between 5-8 (0-9 scale) and the group goes from 1-7.

Just throwing any old compressor or limiter on the stereo mix to the recorder isn't going to sound good. There are only a handful of units out there that could do that without making a mess with a lot of pumping and breathing. One of them is an Aphex Compellor which tends to be rather spendy, but for good reason.

I don't think my ear is trained enough to recognize a good compressor from a bad one, but for now at least I'm using the tt24 built-in compressors. Besides, isn't digital compressing a lot easier to do than analog compressing?

If you're really interested in making this a decent recording, set your gains correctly and use a multi-track recorder (whether to computer or standalone like the Alesis HD24) to record everything - pre EQ/fader/dynamics/etc. But make a reference with your ML off a pair of auxes as a starting point.

We have a couple of tt24s and the lightpipe direct outs are great. You can use that to go into an HD24 (say) which will give you 24 channels @ 44.1/48, or for 32 channels into a computer you can use something like the Presonus Firestudio Lightpipe.

I've done multitrack recordings in the past with my Firestudio Project, and I've considered going this direction, but we're not for 2 reasons.

1. This is a school that lacks an actual class for this, and therefore also lacks the time outside of the concert to dedicate to properly mixing the multitrack recording
2. Learning how to effectively manage PA and recording feeds simultaneously on the fly is something that I feel that I (and any sound guy) should be able to do well.

I will probably get a Presonus Lightpipe (I'm in charge of ordering for the school) so that we have the capability, but in general we just cannot do a full-scale multitrack recording.

One additional thing while I beat this dead horse, on the tt24 you can link 2 auxes to make a stereo aux and then you can use aux mode (which puts the aux levels on the faders) & aux pan button to do set levels/pan on the recording aux, which is about as close as you'll get to easy. That plus the ML would get you something usable, possibly.

That's what I do every concert. Which, incidently, is why my original question had nothing to do with this, but thanks.

 

[FMEng]

. I don't think my ear is trained enough to recognize a good compressor from a bad one, but for now at least I'm using the tt24 built-in compressors. Besides, isn't digital compressing a lot easier to do than analog compressing?

Basic compressors all have similar benefits and tradeoffs, whether implemented in analog or digital. In fact, most digital compressors simply emulate the function of the analog ones, because that is easy to do and familiar to most users.

Dedicated digital audio processors can do sophisticated things that would have been way too complicated to do in analog. There are processors that work with multiband AGC, compression and peak limiters. One example is the Orban 8600, which costs over $10,000. But they are typcially used for broadcasting and mastering.

The ones found in digital consoles are pretty much like their analog cousins because anything else would use too much DSP power needed for other functions.

 

[blalew]

That's what I do every concert. Which, incidently, is why my original question had nothing to do with this, but thanks.

Hilarious. Well, I'm glad I've made it back around to the original question without providing any help

It might be worth it to do a multitrack recording, if only to be able to replay the recording through the board again & find where your problem is. We use the HD24 this way... tt24 -> lightpipe -> HD24 -> 24 ch analog snake -> tt24 (via line). Once you've done the recording, just press all the LINE buttons on each channel & set gain to unity & remix away.

You seem pretty knowledgeable (at least more so than I) but the only other thing I can think of is which cans you're using - need to be closed back. We use Senn HD280. (you probably already know this).

We typically record off a matrix, but our room is rather large (1100 seats) so the mix through the PA translates well. We run that out of the AES output into an Aphex Compellor 320D which as someone mentioned levels things nicely, but your Alesis ML should work reasonably well.

I assume you've compared the pre & post ML tracks to make sure it's not the ML that's causing a problem? Maybe post a sample of what you don't like.

-B

 

[museav]

I may have missed it being suggested, but have you considered adding an ambient mic or two? It is very common for recording or broadcast of live performances to have something to pick up all the ambient sounds that are present if you are actually in the audience. And that is not just the audience, HVAC, etc. noise floor, it is easy to overlook the effect the room has on what is heard with choral or orchestral performances. From early reflections for envelopment to reverb tails, sounds at the listener tend to relate to a longer period and to blend more than you get from a direct recording, particularly when listened to with headphones that add no environment of their own.

A simple example would be if you have a live room then at a listener the attack or leading edge of one note may overlap the reverb tail of a previous note. That may not happen with a close or overhead mic, the first note 'passes by' and then the second note hits it. The result is that the room can affect the dynamics of the music both in timing and level. Some ambient mics may help with incorporating this into your mix.

 

[MisterTim]

You seem pretty knowledgeable (at least more so than I) but the only other thing I can think of is which cans you're using - need to be closed back. We use Senn HD280. (you probably already know this).

I also use HD280s.

I assume you've compared the pre & post ML tracks to make sure it's not the ML that's causing a problem? Maybe post a sample of what you don't like.

While we're talking about the ML, I feel like this is a stupid question, but:

When I'm compressing/limiting, how do I know where to set the threshold? Assuming nothing is clipping, every song is going to be at a slightly different level below 0, and compression/limiting happen before normalization, so how do I know what threshold to compress at?

I may have missed it being suggested, but have you considered adding an ambient mic or two? It is very common for recording or broadcast of live performances to have something to pick up all the ambient sounds that are present if you are actually in the audience. And that is not just the audience, HVAC, etc. noise floor, it is easy to overlook the effect the room has on what is heard with choral or orchestral performances. From early reflections for envelopment to reverb tails, sounds at the listener tend to relate to a longer period and to blend more than you get from a direct recording, particularly when listened to with headphones that add no environment of their own.

I have not, mostly because the HVAC is rather loud. And for a 20-person a cappella group, (correct me if I'm wrong) all I'm going to be getting is PA, which is not really what I want, right? Just for reference, it's a 600-seat house.

For the other groups (orchestra, full choir, etc) that don't need PA reinforcement, I'll have to try that next time.

 

[Morpheus]

I have not, mostly because the HVAC is rather loud. And for a 20-person a cappella group, (correct me if I'm wrong) all I'm going to be getting is PA, which is not really what I want, right? Just for reference, it's a 600-seat house.

For the other groups (orchestra, full choir, etc) that don't need PA reinforcement, I'll have to try that next time.

The point is to record (more or less) what the audience hears... so if the audience only hears a really loud HVAC and PA only... either that's how you want the recording to sound, or you shouldn't even bother mixing for audience and just straight mix for recording.

 

[gpforet]

Should your gain structure be entirely intact, then hard knee limiting should reside somewhere just below unity.

Compression is an entirely different matter and I have never been able to set compression threshold by a number. If I have chosen to insert a compressor than I'm looking for something in particular that I'm trying to achieve (either I'm trying to dampen dynamic peaks or fatten a sound in some way either by parallel compressing or fattening the falloff in a voice). Knowing how a compressor works and using it to alter what we hear are two different things.

When I'm compressing/limiting, how do I know where to set the threshold? Assuming nothing is clipping, every song is going to be at a slightly different level below 0, and compression/limiting happen before normalization, so how do I know what threshold to compress at?

.

 

[blalew]

This is why I was saying to grab it, put it in a DAW & experiment with different levels that sound good. Alternately, record it to the HD on the ML & then 'master' it to CD to see what you like (much more time consuming).

Any way you cut it you'll probably just be getting a 'reference' track, suitable for listening but not enjoyment

Post a sample of the uncompressed & compressed cut if you can.