Mix compression tutorial now uploaded at Samplecraze

Hi guys/guyettes

The latest in the line of compression tutorials mix compression is now uploaded.

This video tutorial deals with all aspects of compressing a mix, in this instance Hip Hop, and goes into detailed explanations of the different types of compressors and their features (variable-mu, FET, OPTO, VCA etc).

Additionally, different forms of compression are applied to the same mix to show and guide how to achieve different compression textures like: dense mix, a transparent mix, a thick and warm mix etc.

I hope these tutorials are helpful and thank you for your continued support.

Zukan

damn, that sounds cool
id take a look if i had time!

nice
thx

Just watching the demo clip the one concern I have about your approach in general is when you say "I look for the peak points". Compression and signal processing is not about looking at anything it's about using your ears and listening.

No, it's also a visual process. We always look and listen. If you rely solely on ears you could clip and still not hear it.

In all aspects of production and dynamic processing we use ears and eyes.

This is why we have accurate metering etc...

Zukan wrote:No, it's also a visual process. We always look and listen. If you rely solely on ears you could clip and still not hear it.

In all aspects of production and dynamic processing we use ears and eyes.

This is why we have accurate metering etc...

is this important when youre not concerned 'breaking' your gear?

Zukan wrote:No, it's also a visual process. We always look and listen. If you rely solely on ears you could clip and still not hear it.

In all aspects of production and dynamic processing we use ears and eyes.

This is why we have accurate metering etc...

If I don't hear it clip - who else will? I work on the basis of if it sounds good, it is good. Obviously this relies on a certain standard of monitoring but the intimation seems to be that you find what you need to fix visually and then do it "I look for the peak points."

I definitely recommend accurate metering, checking your levels going into tape or A/D converter on a VU meter is invaluable so that you know you're leaving enough headroom (or overdriving it to the required level, as the case may be). Clip indicators too, if you hear something crunch and you then think "why is that sounding like that, is something overloading" then you can check your meters to see where it's happening.

However, I don't see looking at a waveform to decide what needs to be done as being in the same ballpark.

If that is the practice you adopt then fair enough.

However, I have yet to meet a single half decent engineer or ME that doesn't work visually as well as aurally.

It simply belies any common sense when you suggest that simply using your ears and ignoring the structure of the waveform visually is equally effective when working with peak values that cannot be determined aurally.

We use tiny incremental values when we mix and engineer and these values cannot be assessed aurally. A boost of 1.6 dB at 6kHz is not something you can do solely by ear. We use peaks to form our referencing when compressing. It is standard practice and one that has been adopted for years. The same applies at every stage of the signal path processing. This is why dynamic processes all carry graphs, metering or some form of referencing. A 2dB cut using a bell shaped EQ at 2.7 kHz in an overall mix context would take a bat to determine the effect aurally.

How you can suggest otherwise, particularly when dealing with such finite measurements, simply suggests poor practice or lack of understanding in how we producers work with small values that can only be detected and referenced visually.

However, if this is how you work then all the best to you.

I provide these tutorials from an engineer's perspective and one who has worked with many labels and artistes and my tutorials are provided using tried and trusted methods.

If the methods do not conform to your practices then ignore them and use your own.
But a healthy debate is always welcomed and I hope my responses will correct any poor practices.

My question to you over and over again, which you don't seem to be answering is that if you can't hear a 2db boost or cut at 2.6khz - why are you doing it? If you can't hear it, who can? And if no one can hear it, what's the point. Music is to be heard not looked at on a screen.

If I put a GML 8200 over the mix and boost 3db at 16khz - I can hear the difference. It's only a little bit of 'air' at the very top but when I switch it in and out, it can be heard. I know that I want that sheen at the top end so I make that boost. I don't get the FFT out and check before and after and make sure that it's definitely pushing 3db. Some tracks I might do it and if there's already a lot going on in that range I might only make it 2db, or it might not sound good at all - but the way I decide is with my ears not with a screen.

I'm a professional engineer too, working on major label projects every day - this is the way that everyone I've ever encountered professionally over the years works.

I'm not talking about mastering engineers, I don't know what those cats do, but recording or mixing everyone is most concerned about what's coming out of the speakers.

Before Pro Tools, even up to say 2002, a lot of projects (even mainstream pop music) were still done to 2" tape working 100% analogue. There were no screens to look at for instructions, you heard what was coming off the tape and you adjusted it until you liked the sound. George Massenburg, Bob Clearmountain, Tom Lord-Alge etc all the biggest and most successful and talented people you can name - all worked the way i describe and it was impossible for them to use your method. They still made incredible sounding mixes. Even mastering engineers - if I mix a track (with my ears or my eyes whatever) to 1/2" tape - how the **** are you going to cut a vinyl from that tape. You will have no screens, no FFTs, no waveforms to look at - what the **** are you gonna do?

You are talking about two entirely different processes and their applications.

2" tape, or any tape, works completely differently to a digital medium, the same as an analogue mixer and a digital counterpart.
You can drive into a tape and have saturation. You can't do that in the digital domain.
You can drive into the headroom of an analogue mixer and have harmonic distortion (depending on the build and design).
You can't do that with tape.
And also, for your information, Studer tapes machines come built with VU metering as did almost all analogue mixers.
If we could drive into tape and the mixer's headroom then why have metering in the first place?
Why not let our ears decide?

The visual aspect, which surprises me that you contradict, provides another useful tool to accurately work on material.

You might not hear a 2 dB boost at 2kHz but the fact that you cannot hear it has precisely dik to do with the dynamics and integrity of the file and how it will finally be rendered.
You fail to realise that a tiny boost in a certain frequency will change the harmonic content of another shared frequency and that can have a huge effect on how the dynamic process behaves even though you cannot hear it.
Using your boost analogy in the 'air band' is standard practice and easily heard.
Working on an audio file with compression involves dealing with peak and low level transients and their integrity, and how they get treated, and are paramount to accurate processing.

I have just finished a new tutorial using compression, side-chaining with the inbuilt filter responses and reshaping the auto-release using it's switchable sub circuit,, and the features of this particular compressor (Sonalksis SV 315 MK2) are entirely dependent on both peak transients and low level transients.

The behaviour and final render are crucially dependent on these transients and they have to be visible as most of the processing is hard to evaluate aurally until the final render. In other words, you need to assess the transients, both peak and low level, to be able to select the appropriate settings. If you relied on ears alone you'd be there all day and have real problems simply because the triggering is now dependent on peak and low level transients.

MEs use peak detection methods all the time.

Now, you say you are a pro engineer working with known labels etc, I am not sure I buy that but let us say you are, then how on earth can you tell anyone that using ears alone, when there is another tool available and one we use all the time particularly in critical processing, is the only way forward.

To me, and maybe I'm wrong, you have a basic 'let your ears decide' thinking but little practical experience in how systems are built and used.
If ears were the only tool we used then we would not need metering, we would not need the names you throw about like Massenburg (who was critical in filter response design), and we sure as hell could do without companies like Sonalksis, Waves, URS and so on who define transient response filters and sub circuit emulation if ears were all we had.

How you could dissuade others from using one of the most critical tools at our disposal defies belief.

Zukan wrote:You are talking about two entirely different processes and their applications.

2" tape, or any tape, works completely differently to a digital medium, the same as an analogue mixer and a digital counterpart.
You can drive into a tape and have saturation. You can't do that in the digital domain.
You can drive into the headroom of an analogue mixer and have harmonic distortion (depending on the build and design).
You can't do that with tape.
And also, for your information, Studer tapes machines come built with VU metering as did almost all analogue mixers.
If we could drive into tape and the mixer's headroom then why have metering in the first place?
Why not let our ears decide?

The visual aspect, which surprises me that you contradict, provides another useful tool to accurately work on material.

You might not hear a 2 dB boost at 2kHz but the fact that you cannot hear it has precisely dik to do with the dynamics and integrity of the file and how it will finally be rendered.
You fail to realise that a tiny boost in a certain frequency will change the harmonic content of another shared frequency and that can have a huge effect on how the dynamic process behaves even though you cannot hear it.
Using your boost analogy in the 'air band' is standard practice and easily heard.
Working on an audio file with compression involves dealing with peak and low level transients and their integrity, and how they get treated, and are paramount to accurate processing.

I have just finished a new tutorial using compression, side-chaining with the inbuilt filter responses and reshaping the auto-release using it's switchable sub circuit,, and the features of this particular compressor (Sonalksis SV 315 MK2) are entirely dependent on both peak transients and low level transients.

The behaviour and final render are crucially dependent on these transients and they have to be visible as most of the processing is hard to evaluate aurally until the final render. In other words, you need to assess the transients, both peak and low level, to be able to select the appropriate settings. If you relied on ears alone you'd be there all day and have real problems simply because the triggering is now dependent on peak and low level transients.

MEs use peak detection methods all the time.

Now, you say you are a pro engineer working with known labels etc, I am not sure I buy that but let us say you are, then how on earth can you tell anyone that using ears alone, when there is another tool available and one we use all the time particularly in critical processing, is the only way forward.

To me, and maybe I'm wrong, you have a basic 'let your ears decide' thinking but little practical experience in how systems are built and used.
If ears were the only tool we used then we would not need metering, we would not need the names you throw about like Massenburg (who was critical in filter response design), and we sure as hell could do without companies like Sonalksis, Waves, URS and so on who define transient response filters and sub circuit emulation if ears were all we had.

How you could dissuade others from using one of the most critical tools at our disposal defies belief.

when working with hardware (no waiting for rendering).. is this looking at meters still important? because your points contesting 'only what you hear is important' seem only valid when in pre-rendering, to me... but maybe im missing something

We have always had metering at every stage of both processing and aligning devices simply because certain processes take place when you enter the headroom of any device or application. Additionally, calibration comes into the equation when dealing with devices that are chained together and metering is critical. When aligning certain equipment we need test criteria and metering or any form of referencing is crucial. In software it is as important. A lot of engineers will use accurate metering and some will even use VU metering weighted against it's digital counterpart.
But in terms of compressor sub circuit designs and filter responses referencing becomes even more important.

You have to accept that ' if it sounds good it is good' is a great analogy but not the only one. When dealing with peak sensitive devices or software it is as important to be able to see the waveform and how it behaves as it is to hear it.

In my compression example, I used the uppermost peak transients as referencing when trying to align the lower transients. This is actually common practice.
This is also very important when setting attack times as some compressor designs like the FETs will act very fast on peak transients and end up compressing cycles instead of peak transients, and for that peak referencing and detection becomes your best friend.
I have gone into this in my mix compression tutorial, how different design compressors behave and how that affects peak transients.

At the end of the day we use as many tools as possible to get the job done professionally and the two most important are eyes and ears.

Paul_Caruso wrote:My question to you over and over again, which you don't seem to be answering is that if you can't hear a 2db boost or cut at 2.6khz - why are you doing it? If you can't hear it, who can? And if no one can hear it, what's the point. Music is to be heard not looked at on a screen.

mixing in a _real_ studio environment with at least half-decent monitoring, you can hear this easy.

You can drive into the headroom of an analogue mixer and have harmonic distortion (depending on the build and design).
You can't do that with tape.

you think the tape deck doesn't have an amp before the actual tape head? come on....


anyhow.... at least how I do mix compression (and most of the other engineers when mixing with hw) you strap the mixbus compressor on the mixbus. no peak detection, no looking at waveforms... nada. you only use your ears, perioid. sometimes when using a hw compressor on stuff not looking at the cotrols, can result in some pretty weird settings. it's all about ears. btw, a real analog VU meter reaction time is slower than maybe a fast peak on a drum. so the VU meter is only for "about what is going on". I wouldn't rely on visual metering either.


my point, there is more than meets the eye.

Eemo1 wrote:

You can drive into the headroom of an analogue mixer and have harmonic distortion (depending on the build and design).
You can't do that with tape.

you think the tape deck doesn't have an amp before the actual tape head? come on....

Different process and it's called saturation. Usually, the decks are driven at the mixer source and it's the opamps that process the signal along with the VU amps.

Eemo1 wrote:
anyhow.... at least how I do mix compression (and most of the other engineers when mixing with hw) you strap the mixbus compressor on the mixbus. no peak detection, no looking at waveforms... nada. you only use your ears, perioid. sometimes when using a hw compressor on stuff not looking at the cotrols, can result in some pretty weird settings. it's all about ears. btw, a real analog VU meter reaction time is slower than maybe a fast peak on a drum. so the VU meter is only for "about what is going on". I wouldn't rely on visual metering either.


my point, there is more than meets the eye.

So, how do you deal with peak and low level transients when dealing with compressor sub circuits and filter response curves?

look, this has become an epic and subjective debate, but you guys seem to be avoiding the most obvious here: that compressor designs are dependent on transient processing and sub circuits are built specifically for this. You also claim that ears are the only thing that counts.
Fine by me.

You work however you want to but if you ask me to explain a process that involves the physics then be prepared to accept that there is both theory and application involved and it is based on not only the process but the design and application of that process. This is the physics and I would be amiss not to include it in my tutorials. Otherwise, it might as well be a turn a knob or two and don't bother with waveform integrity and just go with your ears. In other words, all you need is my primer tutorial on how a compressor works and the rest is down to you.

BUT, you work how you like and obviously my tutorials are not for you.

All the best to you guys and time for me to leave this debate.