Topic: Does dynamic range compression and other effects sound 100% natural?

[soundman2013]

Hi, I'm making a webpage for my community that will have recordings of vocals and many different types of acoustic instruments in an outdoor setting. I am always in the same recording spot, but the location of the performers is constantly changing, as do their volume levels. There is usually ten to fifteen feet between me and the performers, and sometimes some people between us as well. Two musicians have told me that they use dynamic range compression during post-processing to raise the volume levels of their recordings. I did a little research and there seems to be some debate about whether this is a good thing (see http://en.wikipedia.org/wiki/Loudness_war). The thing is that I want my recordings to sound as natural as possible--that is, to reproduce as much as possible the actual experience of hearing these performances live so that they capture the energy and feeling of the performances with all their flaws. Do you think it would sound 100% natural if I use dynamic range compression, or for that matter any other post-processing effect like amplification or normalization in Audacity? I am a little worried that some recordings won't be loud enough, so perhaps it is a trade off between losing a little bit of the natural sound of the recordings vs. having recordings loud enough for one's listeners to enjoy. I have a feeling though that the recordings on the many concert hosting websites out there don't have any post production. Does anyone know about this? By the way, I'm using the internal mics on a Olympus LS-11, and to try to get all my recordings as loud as possible, I put the recording volume at 10 unless there are drums or other loud instruments. My other settings are limiter and low cut switch off, mic sensitivity on low, and 96 khz 24 bit wav. I do lose some recordings due to distortion b/c of the recording volume, which I told the musicians. That is when they mentioned compression, advising me to turn the recording volume down and use compression later on to raise the volume level. But I just don't know if the recordings will sound completely natural with compression. Everyone's input would be greatly appreciated.

[bombdiggity]

Spend some time in the processing threads on this site or other sites that cover basic post processing practices. 

You have to set your recording levels in response to the peak levels at the show.  Do not run them arbitrarily loud and clip things (you seem to have found that out but may not then be responding as you should).  You can absolutely raise the levels of anything later but once something is digitally overrecorded it is almost impossible (or is impossible) to effectively repair.  Leave some "head room" and then boost the output in editing.  Peaks at 3 to 6 dB below 0 is a fair margin (but the exact amount is subjective). 

In terms of raising levels in post you have many choices:

Amplitude (volume) just boosts the entire thing by whatever amount you specify.  It is totally neutral so does not change anything except the perceived volume.  All software should be transparent in doing that (though not sure if it all is). 

Other techniques may change the sound (and certainly the dynamics) from slightly to significantly and the results will vary with the precision of the software. 

Normalizing can be done in several ways.  A simple explanation is here:  http://en.wikipedia.org/wiki/Audio_normalization  In theory a neutral sort of basic normalization will automatically put as much output into your result as it can handle without changing the dynamics (giving you the highest potential output without altering any dynamics).  Other sorts of "normalization" may change the dynamics. 

The loudness war argument is valid, but generally refers to very heavy compression (or as the above wiki note terms it certain techniques of "loudness normalization") where clipping is introduced.  It becomes the equivalent of running your recorder at a volume that is several dB's too high for the program (and then sounds like those lost to distortion). 

There are also various forms of compression.  Nearly all music releases employ varying degrees of compression and often several types together. 

Hard limiting is a good option if you have a few spikes and the rest of the program is more or less in a fairly even range.  A hard limit will boost the rest to a set level but boost the spikes less.  You have to be able to run that in a way where you can preview the result to make sure you're not pushing the spikes over 0 db into clipping (or clipping them at whatever upper limit you set).  It is sort of like the limiting function on the recorder but done consciously in post and hopefully done in a way to ensure nothing clips (whereas the limiters on most recorders are pretty harsh and do not prevent clipping - they just reduce the impact after something clips). 

"Compression" generally refers to techniques that reduce the dynamic range in order to provide room to increase the average output level.  Done properly it can enhance the listening experience. 

There are different compression functions that will have various effects depending on the settings and the software.  Use with care. 

Not all software is equal.  You often get what you pay for (or don't).  Audacity is probably OK for volume adjustments, but at least some of the higher end processing algorithms are not up to the standard of commercial programs. 

The musicians were right that you want to back your levels under the point of clipping and then when processing it afterward add levels in a way that also remains under the point of clipping.   

[soundman2013]

Thanks for your response. Regarding normalization, I was going to post each recording in FLAC, MP3 320 kbps, and MP3 variable bit rate, as well as have an embedded play link for each song, and for the MP3 files I just thought about using MP3Gain (http://mp3gain.sourceforge.net/) rather than doing any post production in Audacity. The replay gain plugin in Winamp seems to work very well. MP3Gain might let me level the volumes without changing the dynamics of the recordings. Also, this is a digression, but the only website that seems to let one have public download links, embed codes for playing each recording, 100 gb of storage space and unlimited bandwidth for under $50 is divshare.com. Does anyone know of other alternatives? The site looks good and the free account tested well, but customer service emails go unanswered.

[OOK]

Without going into a lengthy discussion I will say this "trust your ears".  It was the best advice I was ever given many many years ago.  You will know when it sounds unnatural by your ears...just close your eyes and don't look at settings insted listen for it.  I call it mixing blind.  If it sounds unnatural it is.  We performing any post production/editing I listen very closely with both headphones and studio monitors. I trick it out till it sounds unnatural then slowly bring in back to where it sounds right and things sound balanced.

Peace OOK 

[rjp]

Look at the parallel compression thread for some more ideas. The basic idea is to take a duplicate of the original "dry" signal, apply compression, then mix the resulting "wet" signal at a lower level with the "dry" signal. It's a bit fiddly to use, but it can work very nicely for bumping up the apparent loudness without utterly crushing the dynamic range.

[stevetoney]

Do you mean that from one performance to the next the musicians are changing locations, or within the performance the musicians are changing locations?  The reason I ask is you refer to the loudness wars article which, to me, is an issue of the perceived loudness of a recording on playback rather than just a compression issue.  Engineers/producers use a ton of compression (or so it seems) such that when you look at the waveform in a DAW sofware, it's nearly a solid brick.   But my opinion is that they do this for the sole purpose to increase the perceived volume of a recording (which most on ts.com agree is just ridiculous) with no corresponding distortion...and yeah there can be a huge loss of dynamic range as a result.

If the answer is the latter, I'll defer to others here that use compression because I rarely use it so I'm no expert, but if the issue is the former, I don't think the answer is a compression issue at all, unless you want your recordings to brick like so many commercial recordings do.  In that case, you'd need compression to keep your sources from distorting.

So, if the answer is that the loudness changes from one performance to the next, then all you need to do in post for them to have a consistent loudness is to use the 'normalize' tool and normalize your recordings to 0db.  Doing this retains the complete dynamic range of the master and doesn't change the dynamics of the recording at all...it just changes the perceived playback loudness.  While doing this also raises the volume of the noise floor, most recorders and preamps are quiet enough nowadays that most of us don't perceive an increase in noise when doing this (unless your peaks when recording are way down at -20db or so). 

However, I would finally point out a third solution...which is a sorta hybrid solution that I've used in favor of compression.  There are times when I'm recording a performance with a huge dynamic range, such as perhaps performers that go between acoustic and electric instruments.  The acoustic parts of course are much lower levels on my recording than the electric.  In that situation some people like to use compression, but as you point out, I tend to get results that seem to mess with the sound.  I have greater success when I normalize the loud passages to zero and then isolate the soft passages and raise the levels manually to whereever they need to be to satisfy me that I'll be able to hear them on playback.  It can take more time to do this than invoking the compression tool, but I think it's worth it.  I use the fade in and fade out tool at the beginning and end of the passage to make sure the transitions are smoothe and transparent.  Obviously, if you just select a song and applied gain without smoothing, there will be a step increase and decrease in volume that's noticeable on playback.  The longer the fade in and fade out you can make before the desired passage starts/ends, then less noticeable the transition. 

By the way, I assume you're recording in 24bit...if not you DEFINITELY should be doing so, since the increased audio resolution will improve the results of any post-processing efforts you may wish to perform.

[soundman2013]

Thanks for your replies everyone. The position and loudness of the music changes between performers, but there is only a few seconds' break between performances, so there isn't always time to adjust the recording level. I was thinking of your hybrid solution tonedeaf, so I'll give it a try and see how it works. I thought MP3Gain worked well, but I didn't have the best results recently, so maybe I'll try volume normalization in Audacity. Thanks for letting me know that this effect doesn't change the dynamic range. I really don't like the idea of doing that. And I am recording in 24 bit.

[F.O.Bean]

If ypou just need gain in post, I just Normalize to 0db or "add gain" til Im happy with it. Thats usually the ONLY editing I do

[boyacrobat]

compression sounds un natural to me
it plain sucks for strict 2 ch captures like we do

parallel or linear multi band compression is a muddy cluster f  k
to a trained ear

squash a cake and eat it
or eat it the way its served

some like the mud it creates on 2 ch work
i dont

but record an album with 48 ch to use and compress this or that
and it then can be used as a good tool for sonic shaping

2 ch  compression is a mess on aud captures full stop
just raise the levels to required spot -6 or -4  or lowe and learn to use your volume knob
on amp.

a far better and cleaner choice

this is my opinion

test and listen for yourself and see what you prefer
if you have accepted the loudness war you will prob
want to go down that mess of a road and compete with the vol
that has now been raised at the expence of the whole track

headroom and dynamics is the real world

to get back to proper dynamic range practices someone has
to voice the outrage going on

know im one of the voices trying

good luck and happy captures

g

[Brian Skalinder]

2 ch  compression is a mess on aud captures full stop
just raise the levels to required spot -6 or -4  or lowe and learn to use your volume knob on amp.

a far better and cleaner choice

I think whether or not it's a mess it depends a lot on the degree to which one uses compression.  It needn't be a mess unless it's overdone.  Sometimes simply raising levels doesn't address the issue, and sometimes it's just not practical to reach for the volume knob.

headroom and dynamics is the real world

to get back to proper dynamic range practices someone has
to voice the outrage going on

Applying compression to an aud recording, for example one with substantially louder applause than music, hardly compares to the way and degree compression is used in the loudness wars.  There's nothing outrageous about the former, though I understand people's issues with the latter.