Use of compression in guitar recording

[Tricky Fish]

I’ve never used compression for my (basic) recordings, but have recently been listening to some youtube channels demonstrating the use of compression. I’d like to learn more about it, primarily in the context of jazz chord-melody on acoustic guitar.

There’s a lot of experience here of the forum and I’d like to hear from those that have done a lot of recording of acoustic guitars:

Do you use a compressor? Why?
What advice re use of compression do you have for someone who is a beginner at recording?

[grahambop]

I don’t think I’d use it for solo acoustic or classical guitar, it just doesn’t seem all that necessary somehow. Maybe the combination of mic and acoustic signal naturally tends to even out the sound, I don’t know. But I haven’t done much recording of this type.

Where I do use a small amount is for recording my electric guitar (a Gibson 175), mainly because I record direct and there are usually some peaks in the signal. I even these out a little just to make mixing a bit easier and (for solo guitar) to smooth out any loud notes which can sound a bit jolting to listen to.

[RJVB]

Oh, hey

I think it's going to depend on how you'd like to listen to your recordings, and how good your equipment is.

If you listen through headphones and/or in a quiet setting and your recording equipment is good enough, then you'll probably hear the quietest notes anyway without compression. I do this; I set my gains such that the loudest passages I plan to play give me peaks between -6dB and -3dB (in Audacity). Then, in post-prod, I use the amplify "effect" to normalise to 1dB; either the entire track or a more representative section if there are transient peaks. Clipping a short peak in the recorded signal is rarely every audible.

You could record through a tube preamp with an optolimiter (ART make one). The tube already does a bit of compression from what I understand, and the limiter has a similar, configurable function - on the loud end of things.

Producing background music ('muzac") is a whole different thing

[supersoul]

For solo guitar I would avoid it while recording. You can always compress the recording later, but you can't uncompress it if you don't like it.

[Gabor]

I have the feeling that on the majority of jazz guitar recordings there is a compressor *postprocessing* This is what distinguishes the "raw" amp sound we all know from the way more ballanced sound used to hear on the recordings.

[RJVB]

For solo guitar I would avoid it while recording. You can always compress the recording later, but you can't uncompress it if you don't like it.

True, but it's also very frustrating if you just recorded something nice only to discover you had dialed in your gain too aggressively or conservatively

[vintagelove]

If you're a beginner, stay away from tracking with compression. What you should do is download the best compressor plug in ever, which has a completely free version, it is

Fircomp from JonVaudio


I have tons of plug in comps from the top companies, uad/apogee/etc. The Fircomp is the only one that reminds me of using my high end outboard gear. It's a genuine game changer and pretty much replaced everything else.

[citizenk74]

Whatever you do, preserve the original tracks, if they are the take you want. Processing can happen later, if and when, and to the extent necessary. Archive!

[Gilpy]

Compression is a useful tool for recording that is often misused or overdone. It reduces the dynamic range of the instrument and can smooth out unwanted peaks and dips in volume. It can also suck the life out of your recording if applied too heavily.

For guitar, mild compression can be helpful for fingerstyle playing to even out the response. The sound of an acoustic flat top guitar, heavily compressed, is an important part of "classic rock".

Compression is generally not needed for distorted guitar tones as the distortion itself limits dynamic range. For clean guitar tones, I would stay away from compression except to remedy a problem.

Always try a low ratio first, like 2:1 and only use enough to solve your problem. If you can hear the compression "pumping" it's time to back off. Modern compressors allow you to compress within a specific frequency range, leaving everything else relatively untouched.

[Rene Asologuitar]

My ua volt 276 built in compressor works great.
I am able to achieve quality recordings.
Rene

[joeUK]

I use compression on almost every recording and even tho it is reducing dynamic range - it can breath life into a track. I haven’t tried it on solo guitar, but I’d certainly try it (albeit subtly).
I use (UAD) compressors after recording the track - that way I don’t need to commit to keeping it if doesn’t sound good.
I’d def check out the plug-in Vintagelove mentions …

[VanEpsInDeChirico]

Agree with most of what's already been said. I'll add that compression in recorded music creates a very specific (and to me often pleasant) sort of ambiance. Crudely put, I see it this way: Little to no compression often creates a live room sound. Compression can create an unreal but pleasant illusion that you've got your ear right up against each instrument at once. The snare drum sounds like its 2 inches from your face. The picking hand of the guitarist sounds like its inches from your face. The bass, etc. If applied well, the compression of dynamics brings out timbres and textures from an ensemble recording that can only (sometimes) be heard when a solo instrument is played. In my opinion the 1970s was the pinnacle of using compression right. An example off the top of my head:

[j4zz]

My initial take on compression is that it is the evil behind the loudness wars in the music industry. A lot of casual listeners relate loudness to sounding "better".

With Jazz at least, you shouldn't be trying to be louder, and achieving it via compression takes away from the expressiveness of some playing. I would only use compression when a particular type of sound is desired.

[RJVB]

Compression is not a way to sound louder, it's a way to make everything sound at the same volume.

FWIW, (really) good players already apply some level of compression; they know how to sound like they're playing ppp while really being at mf level (or so). Some instruments actually impose that kind of subterfuge.

[Alter]

I strongly dislike playing with compression unless going for a specific effect. But I use it in recording, compressing just slightly in the daw after I record.

You have to watch some tutorials to understand what each parameter does, or/and just use presets in your daw to see if you like it. It takes practice and attention to detail to get to hear all that, it took for me at least, compression is still the most difficult effect for me to decide upon in recording.

I use it to very slightly compress the dynamic range of a track, so the peaks can be problem free. Also I compress on the master track, even on two stereo guitar recordings, as it makes it sound just a bit more coherent.

But I'm talking very minimal compression, where the effect just kicks in at peaks.

YouTube will compress your signal a lot too, especially if it's a lot far apart from -14lfu. I use Youlean loudness meter, a free plugin, and a limiter plugin, to pseudo-master around that.

[citizenk74]

In general, i prefer limiting over compression. That said, I used multi-band compression to achieve key-board like textures in this tune from UTONIA:

Note: Soundclould is dropping the first beat or so. The full phrase is heard in the first iteration. Grrrr!

[j4zz]

Compression is not a way to sound louder, it's a way to make everything sound at the same volume.

Applying compression to a recording can give the impression that it is louder. The average level is higher, because the compression stops the peaks going into clipping, allowing you to raise the recording levels.

[fep]

YouTube uses normalization not compression.

And yes, compression can make a tune much louder and was the weapon of the loudness wars. It no longer works on streaming sites as they measure LUFS and normalize it down (to -14 Lufs on youtube).

[RJVB]

YouTube uses normalization not compression.

IMHO normalisation can become almost the same thing as compression if you apply it adaptively (with some kind of sliding time window) instead of per song/movement.

Is there a cross-platform filter (e.g. a Nyquist effect) for Audacity that allows to amplify a recording to a certain amount of lufs rather than dB?

[fep]

IMHO normalisation can become almost the same thing as compression if you apply it adaptively (with some kind of sliding time window) instead of per song/movement.

Is there a cross-platform filter (e.g. a Nyquist effect) for Audacity that allows to amplify a recording to a certain amount of lufs rather than dB?

I don't use Audacity, but Reaper will render to a user-controlled amount of LUFS. It's an option in the Render window. It's a time saver.

It appears that Youtube applies to the entire song after filtering out long silent sections (you can't get around their loudness rules by leaving a long silence at the end of your video). A guy did a video to tests what is happening with the YouTube normalization, one of the experiments was white noise that started soft and ramped up to loud by the end and it did it for the whole thing leaving the ending louder than their lufs limits. Here's the video:

[Ukena]

Is there a cross-platform filter (e.g. a Nyquist effect) for Audacity that allows to amplify a recording to a certain amount of lufs rather than dB?

The latest version of Audacity has an effect named Loudness Normalization, which allows one to set the LUFS number.

[RJVB]

The latest version of Audacity has an effect named Loudness Normalization, which allows one to set the LUFS number.

Thanks, turns out my version also has that.

However: this has to do some form of compression:

Original:

Loudness normalisation (-14 LUFS)


A simple "amplify" to 1dB as I usual do will set a gain of 2.0 in this case, causing the peaks around t=3s to reach the maximum. It seems that the loudness normalisation amplifies the quieter sections more than it does the louder sections. Normalising to the default -23LUFS corresponds roughly a gain of a bit less than 0.5 (a gain of 4.8 is require afterwards to put the peaks at 1dB while preserving the dynamic range much better).

[supersoul]

Thanks, turns out my version also has that.

However: this has to do some form of compression:

A simple "amplify" to 1dB as I usual do will set a gain of 2.0 in this case, causing the peaks around t=3s to reach the maximum. It seems that the loudness normalisation amplifies the quieter sections more than it does the louder sections. Normalising to the default -23LUFS corresponds roughly a gain of a bit less than 0.5 (a gain of 4.8 is require afterwards to put the peaks at 1dB while preserving the dynamic range much better).

They both just apply gain.

Normalizing finds the highest peak in an audio file and applies gain to the entire file so that the highest peak is at what you set it to be normalized to (-3 dB for instance). Audacity won't let you normalize to a level where anything clips.

Loudness Normalization measures the apparent loudness of an audio file and adjusts the gain of the entire file to get it to the loudness you set it to (-23 LUFS for instance). Audacity doesn't protect against clipping individual peaks when you are using Loudness Normalization.

In a sense clipping is the most brutal form of limiting/compression. I think that's what's happening in your example.

[fep]

Thanks, turns out my version also has that.

However: this has to do some form of compression:

Original:

Loudness normalisation (-14 LUFS)


A simple "amplify" to 1dB as I usual do will set a gain of 2.0 in this case, causing the peaks around t=3s to reach the maximum. It seems that the loudness normalisation amplifies the quieter sections more than it does the louder sections. Normalising to the default -23LUFS corresponds roughly a gain of a bit less than 0.5 (a gain of 4.8 is require afterwards to put the peaks at 1dB while preserving the dynamic range much better).

I'm assuming those red bits are clipping. I wouldn't normalize to where you get clipping. Is that a solo instrument track, perhaps acoustic guitar? To avoid that, I render to -16 lufs to save time. If there is clipping I'll do it again at -18 lufs.

In my opinion, there is no need to squeeze every bit of loudness you can out of a track. And, you should never have clipping (if you just have a few clips, you can use a limiter so they don't clip if you have to have that much loudness).

Normalization is no different than turning up the volume. It's a courtesy to the end user to have consistent LUFS so they don't have to adjust the volume from track to track. But still, the end user can turn up the volume on his playback device if he wants more volume for the entire album.

Turning the loudness wars on its head, highly dynamic tracks like solo acoustic guitar (with their spikey transients) actually sound louder at the same LUFS as band tracks with many instruments and/or compressed tracks.

[RJVB]

Audacity uses red to indicate a peak that reached the maximum value too, so you cannot know whether actual clipping took place. AFAIK there's none when you use the Amplify effect to scale a recording up to 1dB.

That track is indeed a single acoustic guitar.

The only reason I see to normalise to X Lufs is if/when I plan to upload something to YouTube, to make certain YT doesn't do something inappropriate to my recording. But that's probably for a different life. For myself I just do the simple normalisation to 1dB, just so I don't have touch my volume dial when listening (and can concentrate on sound when comparing 2 recordings). Old habit really; back in my (cassette) tape days I also always recorded at the highest possible level (worked better than Dolby C and certainly better than Dolby B IMHO).