Help me master compression once and for all

[bgrotto]

Originally Posted by mgthefuture

I think the one thing I can't get around my head is how do you go about setting a good threshold. Whats the diff in setting it to 9 and knocking down 3 VS setting it to limit at 6 or 8 and knock down 2?

Where you set the threshold determines what part of the program material the compressor will react to. For example, a lower threshold will react to the program material as a whole, while a higher threshold will react to the transient material. Setting the ratio and time constants will fine-tune these reactions.

Obviously, a low threshold will react to transients if the attack time is too fast, and a high threshold won't do much of anything at all if the attack time is too slow.

Similarly, lower thresholds typically work well with lower ratios, as a higher ratio will cause too much gain reduction and will result in obvious gain reduction artifacts (for better or worse). Conversely, higher thresholds often work well with higher ratios: with less signal going over the threshold, a higher ratio is often needed to achieve the desired gain reduction.

As for using compression in a mastering scenario...if you're doing hip hop, there's little need for it, as there is little dynamic range to control. It's usually a much better approach (by "better" I mean "transparent", by the way) to mix through the compressor (if you're looking for a mixed-through-a-compressor sound) and stick to a limiter for the mastering (by "mastering" I mean "loudness", by the way).

If the program material isn't very dynamic, which would be the case in most hip-hop or contemporary R&B styles, compression isn't going to do much but change the balance and tone of your mix. In other words, if you're looking for an increase in perceived loudness, look somewhere else.

I'd start at the mix stage, and expect only a few dB of limiting on the master. Generally-speaking, 2-3 dB of limiting is all one of my mixes require to achieve a "competitive" loudness. And that figure is NOT reliant on whether I mixed through a compressor or not. Any more processing beyond that (eg - the extensive processing that Chris Lago suggested above) is going to completely change the balance of your mix, which is generally not an advisable way to approach mastering.

Originally Posted by mgthefuture

And the release/attack knobs for BUS compressor and MASTER compression...i always see presets or guides online say set attack to fast, and release to 50ms range....why? What are you trying to get it to REALLY do... I do know it affects the transients in some way or another, but how do u detect that its doing too much to them without printing each time you set it? I can detect pumping, but i cant detect whether or not im a MAXIMUM compression VS barely compressing.

This is a purely aesthetic issue, and one man's overcompressed is another man's just right. I don't think you'll get what you're looking for by reading a web forum. You should mix a tune, compress it a hundred different ways (including no compression at all), and listen to the results in a variety of contexts.

EDIT/ADDENDUM: one thing you might want to listen for is how the attack time affects the bottom end. If you set the attack time too fast, you'll lose (for better or worse) some low end. You can work around this by applying a filter to the compressor's side chain, that is, if you prefer the sound of fast attack times but want to keep the bottom end in tact. You can also use a quick attack to control the low end.

Release times have a similar relationship to the bottom end; if they're too fast, the release time will cause audible distortion. Again, experiment with extremes, and every setting in between, to hear how these settings translate to different listening environments.

[Fred Pearson]

I'm not a big mastering engineer but I personally do two types of compression. One to limit the peaks (high threshold, Low Ratio), and one to compress the whole mix (Low Threshold, Higher Ratio). It seems to work pretty well!

Originally Posted by chrislago

I do not activate the dithering and keep the file as is. Then I put it in Voxengo rBrain Pro and convert to 44.1 and 16 bits.

If you're going from 24-bits and then bouncing down to 16-bits, you should probably add dither instead of just truncating it!

Peace

[Chris Lago]

Originally Posted by bgrotto

Any more processing beyond that (eg - the extensive processing that Chris Lago suggested above) is going to completely change the balance of your mix, which is generally not an advisable way to approach mastering.

Good post.

As far as my processing, it's very relative to the masters I get. Most stuff I get is not perfectly mixed and unfortunately, they want it loud so I give them what they want. Just as long as the client is happy, that's all I care about. I don't always add a compressor, depending how the track sounds.

And to Fred Pearson, check out the Voxengo Rbrain pro, it doesn't truncate, it's a dithering tool. I use it all the time.

[Entrainer]

Originally Posted by bgrotto

As for using compression in a mastering scenario...if you're doing hip hop, there's little need for it, as there is little dynamic range to control. It's usually a much better approach (by "better" I mean "transparent", by the way) to mix through the compressor (if you're looking for a mixed-through-a-compressor sound) and stick to a limiter for the mastering (by "mastering" I mean "loudness", by the way).

If the program material isn't very dynamic, which would be the case in most hip-hop or contemporary R&B styles, compression isn't going to do much but change the balance and tone of your mix. In other words, if you're looking for an increase in perceived loudness, look somewhere else.

This is good wisdom. Spoken better than my attempt, which was referring to compression as a groove tool (only shaping envelopes)... there's really no dynamics to control. But even then, better on a drum buss than the master perhaps.

Compression in mastering may be desirable if you have very fast attack, then sidechain the exact same track some milliseconds ahead... this acts like a look ahead.

Then you can achieve similar results to a limiter, but at less extreme ratios. This can be good for layering two limiter type settings with different release times.

And Chris, I think the information you are posting regarding the transients can be a bit misleading. Compressors can actually make transients appear stronger, especially at the settings you are talking about. To control transients, you need an ultra fast compressor/limiter with a look ahead (or the sidechain trick).

I heard of some MEs using SOME outboard compressors to control transients, but I think that control is occurring from the natural program dependent limiting caused by driving the transformers/tubes on the input/output stage. Not so much from the compression action.

To the OP... the book is only available direct. It's a bit expensive compared to other books, but I've found the information to be invaluable.

[Chris Lago]

Originally Posted by Entrainer

And Chris, I think the information you are posting regarding the transients can be a bit misleading. Compressors can actually make transients appear stronger, especially at the settings you are talking about.

Hey I know, it's not clear, unfortunately or fortunately for me, I'm a lot better at actually doing the work than explaining it. Mastering is a bit harder to explain for me, since I just work on a song until it sounds good... So if a person would ask, why this or that? It's because it sounded good to my ears.

[Karloff70]

Originally Posted by Entrainer

But even then, better on a drum buss than the master perhaps.

Hear Hear....SSL type tool on drum buss + bit of tasty limiting at mastering = Bob's your mother's brother!

Millions of plugs will just suck the tone out of it.

[The Fresh Man]

And to Fred Pearson, check out the Voxengo Rbrain pro, it doesn't truncate, it's a dithering tool. I use it all the time.

I guess you must know what you are talking about here. Research truncation and research dither. It's handy to know for a professional Mastering dude.

[B.A.S.E]

Well, if I were you, I would start by trying to imagine how you want the sound to sound like...

For example, you can have the case of very tom-like/pitchy drums... It can be so pitchty that it sounds a bit thin (the transient sound level is quite high but the volume of this sound decreases too fast in simple words)...

In that case you could use compression to make the whole a bit more even, a bit thicker. Where you will put your threshold will mostly depend on your ear. It can make the lowest levels and the highest ones closer if you decreases it.

The rest is really up to your ears, and monitoring system. Actually once you know the basics, it's about training and the monitoring environment.

As far as Bus compression... usually a light one to ease off some transients and get rid of "too much" depth (what people call "glueing").
Easing off the transients might enable to make the whole louder at the mastering stage (since it will give you more headroom).
Well I say that because I guess you are making your tracks louder yourself.

[Chris Lago]

Originally Posted by The Fresh Man

I guess you must know what you are talking about here. Research truncation and research dither. It's handy to know for a professional Mastering dude.

I know the difference between dithering and truncation. RBrain pro is exactly what I need to dither to 16/44.1. GS likes to be really technical sometimes... I just do what I do, no need to explain myself all the time, it's really a waste of my time for the most part, since language/skill barrier doesn't allow me to explain properly anyway. I just know it sounds good when I use Rbrain pro, the rest is unimportant.

I like to keep my life simple. I master the track. Do I need to make the file for Cd? Then Rbrain pro will do the rest. No headaches, no explaining. A/B, yes it sounds fairly close to the original, good enough for me and for the artist. Done.

Actually, to make everyone's lives easier. Here's a website that explains it better than I ever would: Truncation vs. Dither?

Enjoy!

[gorillainthemix]

Originally Posted by chrislago

I just do what I do, no need to explain myself all the time

*looks at Chris' post count*
...are you sure about that?

[Chris Lago]

Originally Posted by The Dutch Master

*looks at Chris' post count*
...are you sure about that?

[gorillainthemix]

Originally Posted by chrislago

fuuckdfegad

[Chris Lago]

Originally Posted by The Dutch Master

fuuckdfegad

I had a stressful night, leave me alone 1 guy was trying to ruin my dj set, he wanted slower (wedding) type of songs while everyone was dancing their butts off to my choice of songs. I had to live through that Lonestar song... That's a lot of torture for one night!

[lllubi]

Originally Posted by mgthefuture

Alright, i think it's about time I nail this compression problem of mine on the head once and for all now.

I've been watching videos on "mastering", using audition, using sonnox, and other tools, and i pay very close attention to EQ and Compression segments.

Of the years i been producing, I've been pretty comfortable using compressor presets or mixing TO the compressor. But now i'm really trying to get my compression "mastery" So i can take things to another level.

My biggest question is what parameters and objectives do you set out to fix when you use a compressor?

I know its a level/dynamic processor, I know when things reach a threshold, it knocks it down by the ratio.

I think the one thing I can't get around my head is how do you go about setting a good threshold. Whats the diff in setting it to 9 and knocking down 3 VS setting it to limit at 6 or 8 and knock down 2?

And the release/attack knobs for BUS compressor and MASTER compression...i always see presets or guides online say set attack to fast, and release to 50ms range....why? What are you trying to get it to REALLY do... I do know it affects the transients in some way or another, but how do u detect that its doing too much to them without printing each time you set it? I can detect pumping, but i cant detect whether or not im a MAXIMUM compression VS barely compressing.

I do know the difference between clippers, limiters, and compressors...but what are the specefic uses in them, if you can technically use a limiter to solve most of those peaking problems?

Let's begin!

watch Michael Costas Mixing Urban Pop at groove3.com
great video inspired by not so helpful comments on GS

explains everything you will need to know to manage your next level

[The Fresh Man]

I know the difference between dithering and truncation.

Just because you looked it up on a website straight away doesn't mean you knew what it was. You said that the rbrain pro doesn't truncate. If it is downgrading the bit rate it is in fact truncating. Just because many software manufacturers have made plug ins for newb mastering it doesn't mean it is mastering.
I'm glad you make money doing what you love but don't pretend reading things on the net means you know about them. And don't pretend you are a mastering Engineer.

[Chris Lago]

Originally Posted by The Fresh Man

Just because you looked it up on a website straight away doesn't mean you knew what it was. You said that the rbrain pro doesn't truncate. If it is downgrading the bit rate it is in fact truncating. Just because many software manufacturers have made plug ins for newb mastering it doesn't mean it is mastering.
I'm glad you make money doing what you love but don't pretend reading things on the net means you know about them. And don't pretend you are a mastering Engineer.

Thanks, you're no 1002 on here who's been calling me out. If it hasn't stopped me to do what I love after no 1000 or 1001, then no 1002 won't really make a difference. Sorry but you'll have better luck talking to a wall.

[bgrotto]

Originally Posted by chrislago

I know the difference between dithering and truncation. RBrain pro is exactly what I need to dither to 16/44.1. GS likes to be really technical sometimes...

Just to be clear (and I'm not saying you don't know this, Chris), one does NOT "dither" to 44.1. Dither refers to bit depth, not sampling rate.

As for GS being "really technical", well, as a "mastering engineer", that sort of attention to detail is kinda your job. Frankly, any professional engineer of any kind should hold himself to a higher technical standard than the layman, so the GS focus on technicality is not only warranted, IMO, but quite welcome. Hell, I wish the standard were higher still; there's way, WAY too much misinformation on this site...

[bgrotto]

Originally Posted by chrislago

Sorry but you'll have better luck talking to a wall.

Heh...biting my tongue on this one...

[Chris Lago]

Originally Posted by bgrotto

Just to be clear (and I'm not saying you don't know this, Chris), one does NOT "dither" to 44.1. Dither refers to bit depth, not sampling rate.

As for GS being "really technical", well, as a "mastering engineer", that sort of attention to detail is kinda your job. Frankly, any professional engineer of any kind should hold himself to a higher technical standard than the layman, so the GS focus on technicality is not only warranted, IMO, but quite welcome. Hell, I wish the standard were higher still; there's way, WAY too much misinformation on this site...

Ok I know what you mean now, and I understand why some people are getting angry in this thread!

Some of these things are so common to me that I just don't think about explaining... Dithering=24 to 16 bit, and then what you guys are pissed off about: RESAMPLING (96k to 44.1k). This is what I was holding back on right?

Does that clear the confusion? Rbrain pro dithers from 24 to 16 and then Resamples depending on the sample rate of the 2 track. All in all, it's a great tool for dithering and resampling. (By the way, it's flat dithering, no noise shaping with it).

if anyone is interested:

Professional sample rate converter software - r8brain PRO - Voxengo

[bgrotto]

Originally Posted by chrislago

Some of these things are so common to me that I just don't think about explaining... Dithering=24 to 16 bit, and then what you guys are pissed off about: RESAMPLING (96k to 44.1k). This is what I was holding back on right?

First off, for what it's worth, I'm not pissed.

Anyway, to clarify your point: dithering doesn't mean 24 to 16 bit. It means (very simply put) adding noise to an audio signal to toggle the lowest bit to minimize quantization error when going from a higher bit depth to a lower one (whether it's 24 to 16 bits or 32 bits to 24, though at higher bit depths there is some argument as to whether dither is necessary).

The wikipedia page on dither is actually quite well-written, with some good diagrams and charts to explain it.

[Chris Lago]

Originally Posted by bgrotto

First off, for what it's worth, I'm not pissed.

Anyway, to clarify your point: dithering doesn't mean 24 to 16 bit. It means (very simply put) adding noise to an audio signal to toggle the lowest bit to minimize quantization error when going from a higher bit depth to a lower one (whether it's 24 to 16 bits or 32 bits to 24, though at higher bit depths there is some argument as to whether dither is necessary).

The wikipedia page on dither is actually quite well-written, with some good diagrams and charts to explain it.

I know what you're talking about, but what about the RBrain pro? It says no noise shaping, flat dither, is that different?

[bgrotto]

Originally Posted by chrislago

I know what you're talking about, but what about the RBrain pro? It says no noise shaping, flat dither, is that different?

Noise shaping basically EQs the dither signal into a specific frequency range so that it is less audible (usually focusing the dither into the higher regions of the audible spectrum, ie - >20kHz), which in effect increases the apparent signal to noise ratio. I can't remember what "flat dither" looks/sounds like, but I'm sure a quick Google search could turn up some answers.

I don't know anything about the R8brain specifically; I've never used it.

[Chris Lago]

Originally Posted by bgrotto

Noise shaping basically EQs the dither signal into a specific frequency range so that it is less audible (usually focusing the dither into the higher regions of the audible spectrum, ie - >20kHz), which in effect increases the apparent signal to noise ratio. I can't remember what "flat dither" looks/sounds like, but I'm sure a quick Google search could turn up some answers.

I don't know anything about the R8brain specifically; I've never used it.

Out of all the dithering tools that I've used, the RBrain Pro was my favorite sounding of them all. Transparent, doesn't seem to affect the sound too much, works for me. I like to go with "what sounds best". Unfortunately, sometimes a lot of things can be beyond me, music theory being the main one.

[bgrotto]

Originally Posted by chrislago

Unfortunately, sometimes a lot of things can be beyond me, music theory being the main one.

Dither doesn't have anything to do with music theory.

[Lrmusic]

Originally Posted by bgrotto

Dither doesn't have anything to do with music theory.

As we speak, John Cage's great grandson is teaming up with Arnold Schoenberg's great great great grandson for their next piece, "chromatic dither".

[Storyville]

Originally Posted by bgrotto

Noise shaping basically EQs the dither signal into a specific frequency range so that it is less audible (usually focusing the dither into the higher regions of the audible spectrum, ie - >20kHz), which in effect increases the apparent signal to noise ratio. I can't remember what "flat dither" looks/sounds like, but I'm sure a quick Google search could turn up some answers.

I don't know anything about the R8brain specifically; I've never used it.

It sounds like "hollow" white noise. Since we're being technical - and I may be wrong - I was under the impression that dithering injects noise that yields a random quantization error preventing the error from being cyclical.

[illacov]

So this guy walks into a bar right? And he asks the barkeep for a scotch and soda. So while the barkeep is looking for the scotch, he gets to talkin to this other guy thats always there, jibber jabberin his mouth and before you know it, the barkeep forgot all about the scotch and soda man. Sad part is the jibber jabber fella, well hell he never buys anything. He just likes to sit at the bar and run his mouth all day long. Some bartender eh??

Strange how the hip hop forum is slowly beginning to resemble Cheers.



Peace
Illumination

[Karloff70]

.......aaaand all over to that other thread to read all Paul Frindle's detailed explanations on dither.....lol

[RTR]

Originally Posted by bgrotto

Dither doesn't have anything to do with music theory.

Thats the "Fruity Loop" generation for ya!

[J CraQ]

R8Brain is a SRC.. Sample Rate Converter.. That's it.