How to compress this? Getting the low volume up.

[Caine123]

hi guys,

i thought I had understood compressors already a long time but i struggle atm to get this wave looking like i want. normally i compress by ear a lot but this time i have a vocal i want to test to compress it more heavily as the highs and lows volumes are a bit too unbalanced.

My original file looks like this

original.jpg

my compressor settings are:
Threshold -18db, Ratio 4:1, Attack 1s. Release 20s = GR -5dB

proc.jpg

and the compressed files shows it only got +0.1 louder but I wanted it a bit more like +0.3

original with comp.jpg

what should i change? afaik I need to raise the Ratio even higer maybe 6:1 or something?

thx

[Zeisner]

I like to use gentle serial compression on signals with high dynamic range (like vocals). Sounds more natural than driving a single compressor hard.

[Benedict]

Compressors can be used to raise overall level buty it is really not where they best shine.

Transients are what the spikey bits are, snare attacks or consonants etc. The tikky bit at the start of sounds that help define them, pull them through the world around ie the Mix.

Compressors are reactive so they hear what happened then start to do what they are set up to do. This makes them great for reshaping a groove, but less great for hacking. By reducing the latter half of a sound, the overall level does drop which is good, esp for improving dynamic range. Unless the compressor can move effectively instantly ie drop the 96dB you want instantaneously without making a splat sound (waveshaping), it is not really the tool for the job.

Clearly we cannot hear this sound - ya know that pic of sound don't work or even add any real value right - so advice can only be vague so probably wrong for this situation.

Assuming a mix or bus:
Step 1- I would perhaps compress, not to hack things off, but to shape the groove. This will lose a bit of level but that is not the intent as such. The intent being to make it more grooved, more space between the transients and the body of the overall sound. This makes it feel more dramatic. Soooo, you may lose a dB or so BUT you gain impact - perception of presence. It feels louder whilst being a bit softer at the numbers. That is a real win. That is Step 1. Without Step you are pushing ships uphill (ships don't work on hills right).


Step 2 - I forget levels for the while and work on tone with some Saturation (NOT Splaturation) and then EQ. This also goes a huge way to making things more present. Not more sub-sub-sub-Bass but more hearability. More clarity, more space engagement.

Step 3 - I finally look at overall levels and compliance (not hitting the top of the tin -0.3dBfs, whilst having nice averages (0VU or a bit over these LOUD days, esp if the material is a bit weak as LOUD covers some of that). I do that with a tiny bit of Limiting. Instead of a super-fast attack, I let it drop right back so there is less actual crush that makes things dull. That handles the bulk of the overs. NOW I use a Clipper. It is a bit soft, depending on the sort of material and how LOUD it needs to feel as opposed to warm. This rounds/hacks off those peaks whilst actually making the overall piece feel more open and fun.

If you really read that with an open-heart - for understanding, not recipes - you should be starting to feel that getting your volume up is not about pushing numbers anywhere near as much as people say and it is about building the perception of presence and excitement that matches the material ie no rigid formula but a mindset.

[rj0]

Sounds a bit like a mild audio leveler application, set for minimal transient impact (transient retention)?

[lfm]

Attack: 1s, that means all transients go by and compressor does not kick in until later.

Like rj0 said, in other words....

[legendCNCD]

@ 7m10s

There are more modes you could try. Levelling compressor/limiter in the style of LA2A could work. I really like TDR NovaGE and KotelnikovGE for these works, but also their Limiter 6 has leveller style compressor.

[kritikon]

As already mentioned, attack = 1 second, which seems hugely long for vocals. Think...how long does it take to say/sing one word? You need to be working in mSecs, not seconds. Unless you're singing in Welsh, most of everything sung is going through that compressor unaffected or very little effect. Maybe sing more about places with long names that take over a second to say, such as llanfairpwllgwyngyllgogerychwyrndrobwllllantysiliogogogoch and your compressor will work .

Also, though I'm not familiar with Fabfilter comp (I generally hate their GUIs) - it looks like autogain is off, and I don't see make up gain anywhere. So you're compressing (very little because it's attack is so long) and then not making up gain. So your whole thing is actually quieter, which is what compressors do if you don't make up the gain. Maybe I'm just missing something on that horrible Fabfilter cartoon-like GUI.

[lfm]

Maybe sing more about places with long names that take over a second to say, such as llanfairpwllgwyngyllgogerychwyrndrobwllllantysiliogogogoch and your compressor will work .

Puts a lot of pressure on the lyrics writer, knowledgable in Welsh geography and think German would be good too, they just add a lot of words together like that.

[billinder33]

The key to good mix-forward, level-balanced vocals is usually small adjustments across several processes.

If you have analog compression available at that tracking stage, IMO that's ideal. An 1176 to shave peaks into an optical comp to do general leveling is a classic combination. Keep it light, don't overcompress.

Once in the box, use light fader automation to smooth out mic distance and volume control issues. Then a light layer of compression on the track for some final smoothing. If you're really pushing for loudness against a hot mix, a clipper on top can help avoid peak overs.

Finally, routing to a parallel compression track will give the vocal a final 'thickening'. Fast attack, fast release, high ratio, but very low volume on this parallel track, to provide a supportive 'floor' for the softer parts.

Sounds like a lot, but again, the 'trick' here is to keep it light at each stage to not completely obliterate the natural dynamics of the vocal. You're micro-adjusting all the various aspects of the vocal's dynamics while preserving the performance as a cohesive whole.

[Funky40]

as has been posted: upwards compressing ( but: i don´t deal with vocals, so.....)


this tool here is the nonplusultra for any such "problems". The uses it can serve for are many. It´s an unbelivably good tool

https://playfair-audio.com/#

there is also a free version (Body) which has only the middle section alive, iirc. Which might even do vs. the OPs case. Totally under-hyped

[Funky40]

Attack 1s. Release 20s

judging from the picture, i would guess the OP meant ms, not Seconds....?

[kernaudioio]

the attack is almost certainly the whole issue. 1 second means the compressor barely fires before the vocal transient is already gone.

try 10-30ms attack, 100-200ms release, ratio around 3:1, then use makeup gain to bring level back up. if the quieter parts still aren't sitting right after that, a bit of gentle volume automation will do more than pushing the ratio higher.

[TIMT]

The disparity in crest factor there is too large for any one tool to address which is why it has been pointed out a few times, parts like this usually require multiple approaches.

Volume automation and comping the takes should be you first port of call for vox, then a leveller e.g LA2A/3A, opto comp of your choosing to bring up the body without necessarily sucking out the life of the performance which is what upwards compression will do if you aren't careful, and if you need more sustain beyond that, best to start looking at paralllel saturation or compression if you are just trying to bring the performance in line and not actually alter it

A decent early reflection will do a lot to massage out some of that perception of level difference too by adding some contrast if the take is mostly closely miked

You can use a fast comp if you want but i haven't found many truly fast plugin comps that don't just eat into the onsets in a bad way due to smoothing on the control signal side including most of the 1176 models (not to mention i just abhor the sound of the 1176 as well, lol). This is most likely a preference thing though, so you do you, but i'd probably just go for another pass of volume automation for the most obtrusive of peaks