Is compression in post necessary on solo acoustic fingerstyle recordings?

[Brent Hahn]

Some of it, for me, is contextual. If it's a solo piece and played by a better fingerpicker than me, it can sound fine with no compression. Or maybe a bit of DIY-Mastering bus compression to goose up the overall average volume.

But if it's in an ensemble/band setting, there's a good chance that there won't be as much evenness from finger to finger and/or string to string as you'd want. Compression can fix that. And the right amount might be surprising. You get it to where it sounds right in the mix, and then solo it, and think "Oh dear me!" and then back off the compression. The point being... don't solo it.

[AcousticDreams]

I suspect the answer is dependent on the players style and the dynamic range in which he plays.

I have heard so many people claim that compression of acoustic guitar should only be applied after tracked. But I have an itch of curiosity, that needs to be scratched when it comes to compression. Sometimes there is no getting around our own personal curiosity. So I have ordered a brand new model from Buzz audio called the Velox. In fact, I believe I will be the first one to get this new model. While it is an opto compressor, it claim is to be still adjustable for attack and release and is also suppose to be super fast. Further more it has a built in Mix control for parallel tracking.

I did hear one engineer on Gearslutz say that people play differently when they hear the source played with compression on tracking. He was referring to drums, but this does make some sense to me. Remembering the past days when I use to track vocals with compression, it actually helped in the way I sung. I started to tame my voice.

Going back to my intro statement of depending on players style; I am reminded of a time when my Best friend came over and played my two best guitars. One was made from Amazon Rosewood and the other Austrian Walnut. when I played played the two guitars in front of him, the Amazon Rosewood was his favorite by far for myself. When he played the two guitars, the Austrian Walnut favored his style of playing...again by far. Further....when I miced him up, he accidentally moved his chair which put the mic in front of the sound hole. Should not have sounded good, But it did. And further it sounded like the guitar. He is a gentle player.

Anyway...my point is...what works for one person may not work for another. There are of course generalities, but those generalities don't work for everyone.

[Doug Young]

Originally Posted by Knives&Guitars

I suspect the answer is dependent on the players style and the dynamic range in which he plays.

For sure, and also the sound/audience you're aiming for. You need to use compression, unless you don't. Or maybe you don't unless you do :-). I just did a quick survey of a few different released CDs, just solo fingerstyle guitar, and compression and levels are all over the map, but do seem to vary by style. Check out these waveforms:

A bit of an Al Petteway tune (about -15 LUFS):

Al.jpg

A Tommy Emmanuel tune (about -14 LUFS)

tommy.jpg

A Calum Graham (Candyrat) tune. (About -11 LUFS)

Calum.jpg

An Antoine DuFour tune (around -8 LUFS)

Antoine.jpg

[rick-slo]

Originally Posted by Doug Young

For sure, and also the sound/audience you're aiming for. You need to use compression, unless you don't. Or maybe you don't unless you do :-). I just did a quick survey of a few different released CDs, just solo fingerstyle guitar, and compression and levels are all over the map, but do seem to vary by style. Check out these waveforms:

At those resolutions somewhat difficult to tell note densities apart from compression effects regarding the filled in solid areas. Dynamic ranges seem reasonably comparable.

[Doug Young]

Originally Posted by rick-slo

At those resolutions somewhat difficult to tell note densities apart from compression effects regarding the filled in solid areas. Dynamic ranges seem reasonably comparable.

They do? Maybe it's not coming thru in the screen shots. The 1st two have lots of dynamic range. Lots of transients, and the only things coming close to 0 db are a few small transients. The final one is virtually a slammed solid block. Listening to them, the last one is roughly twice as loud as the first (7 db difference in average level).

Basically, if you have 2 similar tunes (solo guitar), both with peaks of 0db, but one is 7 db louder than the other, it's almost certainly substantial compression

[lkingston]

I like to play live with no compression. It is trickier at first, but in the long run you have more dynamic range and control over it. I like to duck my playing under my voice and have enough room to play fills, melodies, and solo sections between the vocal sections. You can't do this with anything but the most minimal compression.

[FrankHudson]

Originally Posted by jim1960

Unless what you're releasing is going to cd or is club music, there is no need to engage in the search for loudness. The popular online streaming services all normalize to whatever level they've chosen (usually somewhere around -14 LUFS) and your music won't be louder or softer than anyone else's because of that.

When I master something for a client, the first thing I always ask them is whether the tracks are ever going to be released on cd. If the answer is no, I explain to them what the online services are going to do with their music if I master it to cd levels. For cds, I shoot for the -9 to -8 LUFS range. But the louder you push the master, the more dynamic range is sacrificed. There are some people who master to -7 and even -6 but I think that's mostly club music where no one gives a rat's hind quarters about dynamic range. If the client isn't going to cd with the tracks, I explain that I''ll master to the -13 to -12 range and that's still a little louder than anyone will hear it on a streaming service.

For the most part, streaming has killed the loudness wars.

Jim, to my understanding you're more knowledgeable and skilled than I am, so this is not a debate challenge, but more at wondering how this fits with what I think I know, which could be wrong.

My understanding of simple normalization* is that it takes the peaks to some db level minus from zero (I'm speaking in the digital domain, in the analog to tape world people pushed stuff past zero for peaks on meters when mastering or even tracking, of course producing compression and desired saturation of a type). In music with great dynamics, loud and delicate soft passages within a piece, or from piece to piece in a collection, something normalized in this way will have softer parts that will not be audible in many listening situations (cars, small speakers at lower volumes in non-ideal listening spaces, headphones without noise-cancellation in public places). Some orchestral pieces on record are unlistenable in many casual situations like those for me. They sound great in a quiet room, and probably make listeners wanting just a whiff of background music they aren't really listening to happy enough, but I want to hear the quiet parts.

So, to my not-great skills, compression, including the sophisticated compression modern tools allow and I only partway understand, is a necessary mitigation tactic.

I understand this sort of issue goes back to the radio days (long before YouTube) when most pop music stations compressed the records they played (not just normalization as I understand the term in modern digital tools) -- yet the great pop records of my youth were compressed too before they reached the radio stations, they didn't just say "well, the radio station will compress them for us, so we shouldn't or need not, compress them."

*Is this confusion on my part down to you using normalization to mean something different to the normalization feature as I see in digital domain DAWS. I mean, when things on a collection of cuts are made to sound "the same/or similar volumes" with whatever tools are available, including compression, you are in a dictionary way "normalizing" the perceived levels.

[FrankHudson]

Originally Posted by Doug Young

They do? Maybe it's not coming thru in the screen shots. The 1st two have lots of dynamic range. Lots of transients, and the only things coming close to 0 db are a few small transients. The final one is virtually a slammed solid block. Listening to them, the last one is roughly twice as loud as the first (7 db difference in average level).

Basically, if you have 2 similar tunes (solo guitar), both with peaks of 0db, but one is 7 db louder than the other, it's almost certainly substantial compression

Let me thank you for those illustrative screen grabs. They were very noticeably different to me, indicating what you said they indicated. That last one was a real "stuffing sausage filling into a casing" deal!

[rick-slo]

Originally Posted by Doug Young

They do? Maybe it's not coming thru in the screen shots. The 1st two have lots of dynamic range. Lots of transients, and the only things coming close to 0 db are a few small transients. The final one is virtually a slammed solid block. Listening to them, the last one is roughly twice as loud as the first (7 db difference in average level).

Basically, if you have 2 similar tunes (solo guitar), both with peaks of 0db, but one is 7 db louder than the other, it's almost certainly substantial compression

You mean average volume I guess. Don't know what particular pieces those were. Some pieces have more things going on simultaneously, some are more sparse. Can take more magnification to separate out individual notes and transients.

[jim1960]

Originally Posted by FrankHudson

So, to my not-great skills, compression, including the sophisticated compression modern tools allow and I only partway understand, is a necessary mitigation tactic.

Sometimes, yes. If the dynamic range of a piece (the difference between the loudest and the least audible) is too great, it can leave listeners riding the volume knob in order to have a good listening experience. One of the functions of compression is to reduce that dynamic range, mitigating the necessity to ride the knob.

Originally Posted by FrankHudson

I understand this sort of issue goes back to the radio days (long before YouTube) when most pop music stations compressed the records they played (not just normalization as I understand the term in modern digital tools) -- yet the great pop records of my youth were compressed too before they reached the radio stations, they didn't just say "well, the radio station will compress them for us, so we shouldn't or need not, compress them."

That's certainly possible but I don't have any knowledge of those facts. They may have employed compression or it may have simply been someone riding the fader at the station. Either is plausible and I wouldn't be surprised if both were true depending on the studio's budget and whether they could afford a compressor.

Originally Posted by FrankHudson

*Is this confusion on my part down to you using normalization to mean something different to the normalization feature as I see in digital domain DAWS. I mean, when things on a collection of cuts are made to sound "the same/or similar volumes" with whatever tools are available, including compression, you are in a dictionary way "normalizing" the perceived levels.

I'm not perceiving a difference in the definition of the word as it applies to DAWs or online streaming. In both cases, tracks are being adjusted to meet a predetermined level. The only difference is that when we normalize on a DAW, we're left with something louder than, for example, a track normalized for Spotify. In both cases we're shooting for a universal constant within that realm (for lack of a better word), the only difference being the loudness level we normalize to.
Does that make sense?

[Bob Womack]

Originally Posted by rick-slo

At those resolutions somewhat difficult to tell note densities apart from compression effects regarding the filled in solid areas. Dynamic ranges seem reasonably comparable.

Originally Posted by Doug Young

They do? Maybe it's not coming thru in the screen shots. The 1st two have lots of dynamic range. Lots of transients, and the only things coming close to 0 db are a few small transients. The final one is virtually a slammed solid block. Listening to them, the last one is roughly twice as loud as the first (7 db difference in average level).

Basically, if you have 2 similar tunes (solo guitar), both with peaks of 0db, but one is 7 db louder than the other, it's almost certainly substantial compression

Nice examples, Doug. The screen shots rather clearly show me a range of recordings that differ in the amount of high-threshold limiting applied, progressively from top to bottom. High-threshold limiting differs from compression in that the effect threshold is usually moved above the body of the notes. With a fast rise and fall rate or automatic detection and envelope adjustment, only the transients are affected. Meanwhile, the body of the notes sound virtually unchanged. When gain makeup is applied, the signal can be raised in volume to fill the headroom vacated when the transients were reduced.

The most heavily limited record I've heard was Joe Walsh's Analog Man. I haven't measured the density but they appear to have used brick wall limiting to get it down to the -6 to -4 range.

When I am mastering I typically suggest to clients that we go no lower than -11lufs and usually shoot for between -12 and -14. Below that the kick drum stops jiggling the guts no matter how loud you play it. It won't win any loudness war but it is FAR more musical and FAR less fatiguing.



By the way, in order to accomplish high threshold limiting in mastering I usually apply the UAD Precision Limiter plug-in, a look-ahead brick wall limiter. It works a little backwards from musical limiters - rather than setting the threshold, you set the input level and the output level and move the signal up into the limiting. It offers a real time compression tracking display that indicates how many dbs of compression have been applied. I place iZotope insight in the chain after this limiter to analyze the signal.



Bob

[Doug Young]

Originally Posted by rick-slo

You mean average volume I guess. Don't know what particular pieces those were. Some pieces have more things going on simultaneously, some are more sparse. Can take more magnification to separate out individual notes and transients.

Yes, it'd be more illustrative to show the same piece uncompressed vs compressed, but that wasn't my point. The point was that different commercial releases have varying LKFS/LUFS values. All these pieces were just basically fingerstyle guitar, none particularly more sparse than others. I just grabbed a few at random to check them. You can try it yourself by comparing different tunes and dial in with any degree of magnification you want - tho it's the overall behavior that's more interesting.

The main reason I looked at this was Jim's comment about mastering to -8 to -9. That sounds like serious loudness war levels to me, and I wondered if I was mistaken about what levels the type of music I generally work with was at. As I expected, even the Candyrat stuff which seems really loud to me doesn't hit -8 or -9.

LKFS is generally used as an overall loudness measure - I'm not talking about momentary here. It's the overall (i.e. average - tho the standard is weighted in various ways) loudness. In theory, CD level is supposedly -15 LKFS, at least according to the various settings on the meters I use - I see differing "standards" online, but the streaming services seem to have settled on -14, slightly louder. Meanwhile the Candyrat releases seem to be much louder, probably reflecting a younger set of acoustic players, many of whom seem to have a metal background.

[jim1960]

Originally Posted by Doug Young

The main reason I looked at this was Jim's comment about mastering to -8 to -9. That sounds like serious loudness war levels to me, and I wondered if I was mistaken about what levels the type of music I generally work with was at. As I expected, even the Candyrat stuff which seems really loud to me doesn't hit -8 or -9.

Admittedly, -8 and -9 is at the top of the range but that's what being allowed on a couple of platforms as of Dec 2020. I think -12 would be the better range, and that 's always my target for non-cd material, but there's has been no standardization of any level for anything in the industry and given their options, clients almost always want to push a little louder. The range between -8.0 and -9.9 can be achieved without totally squeezing the life out of a track. There is a loss of some dynamic range but it's not over the top and, frankly, it's what many are used to hearing (and I'm not talking about the solo acoustic or singer/songwriter genres because most of the mastering work I've done has been outside of that bubble due to the mix engineers I work with not working primarily in those areas).


This chart can be found on the Mastering the Mix website.

[KevWind]

Originally Posted by FrankHudson

My understanding of simple normalization* is that it takes the peaks to some db level minus from zero (I'm speaking in the digital domain, in the analog to tape world people pushed stuff past zero for peaks on meters when mastering or even tracking, of course producing compression and desired saturation of a type). In music with great dynamics, loud and delicate soft passages within a piece, or from piece to piece in a collection, something normalized in this way will have softer parts that will not be audible in many listening situations (cars, small speakers at lower volumes in non-ideal listening spaces, headphones without noise-cancellation in public places). Some orchestral pieces on record are unlistenable in many casual situations like those for me. They sound great in a quiet room, and probably make listeners wanting just a whiff of background music they aren't really listening to happy enough, but I want to hear the quiet parts.

So, to my not-great skills, compression, including the sophisticated compression modern tools allow and I only partway understand, is a necessary mitigation tactic.

I understand this sort of issue goes back to the radio days (long before YouTube) when most pop music stations compressed the records they played (not just normalization as I understand the term in modern digital tools) -- yet the great pop records of my youth were compressed too before they reached the radio stations, they didn't just say "well, the radio station will compress them for us, so we shouldn't or need not, compress them."

*Is this confusion on my part down to you using normalization to mean something different to the normalization feature as I see in digital domain DAWS. I mean, when things on a collection of cuts are made to sound "the same/or similar volumes" with whatever tools are available, including compression, you are in a dictionary way "normalizing" the perceived levels.

I am by no means a pro engineer nor have any formal degree , so my understanding is an amateur enthusiast only .
But as I understand it and as per the linked article below
There are two types of normalization One is peak normalization (which I think of as brick wall limiting) and one is loudness normalization (which I think of as compression often with gain) .

https://www.thebroadcastbridge.com/c...lization-works

[Doug Young]

Originally Posted by jim1960

This chart can be found on the Mastering the Mix website.

Interesting chart. No agreement between platforms :-) What makes it harder is that it seems to be a moving target as well. For example, Spotify itself says they will set it to -14, period. Tho then they go on to say that premium users of their app can select anywhere from -11 to -23. (Which seems weird, don't end users have a pesky thing called a volume control? Why would end users be specifying a target mastering level?)

https://artists.spotify.com/help/art...-normalization

I get the difference in genres. The OP was asking about solo fingerstyle guitar, not even singer/songwriter stuff, let alone bands.

I just pulled up one of my son's tracks for a different genre comparison. Punk rock stuff. -6 LKFS! Peaks and transients? Who needs peaks and transients when you can blast them with a solid wall of distorted guitars? :-)

Screen Shot 2021-05-02 at 12.19.04 PM.jpg