- KVRer
- 21 posts since 30 Aug, 2005, from Finland, Tuusula

suup

So I have this problem that and I don't quite understand how it is possible to maximize loudness in the way that you don't hear distortion. Examples of one of this type of track is from Ferry Corsten's remix of Bobina invisible touch or Anahera

Invisible Touch remix:

https://www.youtube.com/watch?v=POGLKoYU5g4

Anahera:

https://www.youtube.com/watch?v=7ZMZHbAKvGA

So if you own either of these songs in wav for you will notice that from the start the peak is at 0 db with the beat and there is no audible distortion that I can hear at least. So how it's possible to master your track this way since what happens when one makes tracks is that audio adds up when you have more sounds in the track. Easy example is plain kick then clap on top of it and when the clap hit the master volume goes up and so on and so on. So please help me understand what the f**k is going on here how it's possible to maximize the over all volume to 0db without audible distortion since I can't do that. I get about at -3 difference with the drums playing, but when bassline and sounds hit in I can't / know how to keep the volume building up

So I have this problem that and I don't quite understand how it is possible to maximize loudness in the way that you don't hear distortion. Examples of one of this type of track is from Ferry Corsten's remix of Bobina invisible touch or Anahera

Invisible Touch remix:

https://www.youtube.com/watch?v=POGLKoYU5g4

Anahera:

https://www.youtube.com/watch?v=7ZMZHbAKvGA

So if you own either of these songs in wav for you will notice that from the start the peak is at 0 db with the beat and there is no audible distortion that I can hear at least. So how it's possible to master your track this way since what happens when one makes tracks is that audio adds up when you have more sounds in the track. Easy example is plain kick then clap on top of it and when the clap hit the master volume goes up and so on and so on. So please help me understand what the f**k is going on here how it's possible to maximize the over all volume to 0db without audible distortion since I can't do that. I get about at -3 difference with the drums playing, but when bassline and sounds hit in I can't / know how to keep the volume building up

- KVRAF
- 2096 posts since 11 May, 2011, from The marxist paradise that is the UK

Reefius wrote:Proper gain staging/mixing and a limiter on the master track should do the trick.

... Along with 2 or 3 compressors on most channels, mico-editing of the audio, not forgetting side-chaining the kick. Details, unfortunately, are specific to the track.

Oh, and 2 or 3 limiters on the master.

- KVRist
- 130 posts since 9 Mar, 2018, from Frankfurt am Main, Germany

Such type of music is really laborious to mix and make manually, since it behaves in a very unnatural way. The funny thing is that it all can be done by AI.

Try using a different Limiter with more options.

Try using a different Limiter with more options.

For DISCOGRAPHY, see К Ɱ Ԏ Ꮇ Ꮩ Ꭶ Ꭵ Ꮳ

- KVRer
- 21 posts since 30 Aug, 2005, from Finland, Tuusula

so far I'm very dissapointed of your answers there's no information on them. Gain staging is not an answers since volume grows when you have more sounds playing same time so that's no brainer. The problem is the overall sound not individual sound. I also compress the sounds separately, but this still is not enough so no help there. Why for example 2 limiters make any difference to 1 when you just crush it more anyway ? would be nice to hear real life examples instead of vaguely talking something.

In my track the difference before bassline and sounds come in after mastering is only 3 db so maby it could be that more "micro-editin" is needed to level out the sounds as a whole and after that it's possible to peak the whole f**king thing. Just annoyed that I don't find any kind of real tutorial about the issue anywhere thought that perhaps here people would know things :/

In my track the difference before bassline and sounds come in after mastering is only 3 db so maby it could be that more "micro-editin" is needed to level out the sounds as a whole and after that it's possible to peak the whole f**king thing. Just annoyed that I don't find any kind of real tutorial about the issue anywhere thought that perhaps here people would know things :/

- KVRian
- 1304 posts since 14 Feb, 2006, from Berkeley, CA

Uhh...people gave you meaningful responses with actual information: gain staging, sidechaining, multiple layers of compression/limiting on tracks & busses, micro-editing. If you don't find those suggestions useful that's a reflection on you.

Gain staging means knowing what/where your highest level will be when kick, clap & bass etc. are hitting at once and producing around that so you get a full sound with no unwanted distortion, while preserving dynamic impact. Look at mastering resources by everyone from Bob Katz to deadmau5 and they will hammer home the importance of these fundamentals.

Gain staging means knowing what/where your highest level will be when kick, clap & bass etc. are hitting at once and producing around that so you get a full sound with no unwanted distortion, while preserving dynamic impact. Look at mastering resources by everyone from Bob Katz to deadmau5 and they will hammer home the importance of these fundamentals.

- KVRer
- 21 posts since 30 Aug, 2005, from Finland, Tuusula

they are meaningful no doubt, but not really helping in technical point of view thus they are only vague road signs that don't help I already know these stuff that's why. I was watching video about gaing staging last night and in it it was only to make sure that mixer track doesn't clip. I will check those ones you suggestested that's more helpful thanks.

- KVRAF
- 11855 posts since 7 Dec, 2004

The amplitude of A + B depends upon the correlation of the signals A and B.

edit: I'm using "50% correlation" here to describe the fact that an entirely uncorrelated signal like white noise when mixed will produce a signal level of sqrt(2). There is no such thing as "less than completely uncorrelated" until you consider discrete cases like a pair of sampled pulses (a single sample value) with different unitary phase (AKA: different samples.) In such cases "mixing" these two signals produces a result with amplitude exactly equal to the greater of the two discrete inputs. It is also possible to be dealing with other discrete impulses where the mixed signal could be anywhere in the range of 1 to sqrt(2) based upon the phase of the source signals. That said, technically speaking; 0% correlation = sqrt(2).

With 0% correlation = 1 (0 dB)

With 50% correlation = sqrt(2) (+3.01 dB)

With 100% correlation = 2 (+6.02 dB)

100% correlation means A = B, they are the same signal.

When you multiply two signals A * B, the result is that the harmonic "bandwidth" of the signals are added. So assuming full bandwidth = 1, where A and B are both full bandwidth the resulting bandwidth is then 2.

Since multiplication of an envelope signal with any other signal is just another multiplication of signals, this rule also applies to compressors, limiters or even volume ramps!

The bandwidth of the envelope signal is added to the bandwidth of the signal being modulated. This is also known as "distortion": harmonics are added!

The harmonics that result from a frequency present in two signals multiplied together produces two output frequencies equal to the sum and difference of the input frequencies at half the amplitude.

Assuming signals A, B both have bandwidth and frequency = 1:

A * B = 1 + 1 and 1 - 1.

That is frequencies: 2 and 0.

Let's say we have a sine input on signal A at 1000 Hz.

Signal B is equal to signal A squared and integrated (calculus) to form a very rough envelope signal.

B = integrate(A^2)

Since A * A (that is: A^2) produces 2x the bandwidth of A we now have a signal containing both frequencies 0 Hz (a DC offset) and 2000 Hz.

The amplitude is 1/2 and the phase matches the input from signal A.

DC = 1/2

2000 Hz = 1/2

So if our input amplitude was 0 dB (-1 to +1) our output is now twice the frequency from 0 to +1, or positive only due to the DC offset from the 0 Hz "difference" signal.

To be continued...

edit: I'm using "50% correlation" here to describe the fact that an entirely uncorrelated signal like white noise when mixed will produce a signal level of sqrt(2). There is no such thing as "less than completely uncorrelated" until you consider discrete cases like a pair of sampled pulses (a single sample value) with different unitary phase (AKA: different samples.) In such cases "mixing" these two signals produces a result with amplitude exactly equal to the greater of the two discrete inputs. It is also possible to be dealing with other discrete impulses where the mixed signal could be anywhere in the range of 1 to sqrt(2) based upon the phase of the source signals. That said, technically speaking; 0% correlation = sqrt(2).

With 0% correlation = 1 (0 dB)

With 50% correlation = sqrt(2) (+3.01 dB)

With 100% correlation = 2 (+6.02 dB)

100% correlation means A = B, they are the same signal.

When you multiply two signals A * B, the result is that the harmonic "bandwidth" of the signals are added. So assuming full bandwidth = 1, where A and B are both full bandwidth the resulting bandwidth is then 2.

Since multiplication of an envelope signal with any other signal is just another multiplication of signals, this rule also applies to compressors, limiters or even volume ramps!

The bandwidth of the envelope signal is added to the bandwidth of the signal being modulated. This is also known as "distortion": harmonics are added!

The harmonics that result from a frequency present in two signals multiplied together produces two output frequencies equal to the sum and difference of the input frequencies at half the amplitude.

Assuming signals A, B both have bandwidth and frequency = 1:

A * B = 1 + 1 and 1 - 1.

That is frequencies: 2 and 0.

Let's say we have a sine input on signal A at 1000 Hz.

Signal B is equal to signal A squared and integrated (calculus) to form a very rough envelope signal.

B = integrate(A^2)

Since A * A (that is: A^2) produces 2x the bandwidth of A we now have a signal containing both frequencies 0 Hz (a DC offset) and 2000 Hz.

The amplitude is 1/2 and the phase matches the input from signal A.

DC = 1/2

2000 Hz = 1/2

So if our input amplitude was 0 dB (-1 to +1) our output is now twice the frequency from 0 to +1, or positive only due to the DC offset from the 0 Hz "difference" signal.

To be continued...

Last edited by aciddose on Fri Jun 01, 2018 9:39 pm, edited 1 time in total.

Free plug-ins for Windows, MacOS and Linux. Xhip Synthesizer v8.0 and Xhip Effects Bundle v6.7.

- KVRAF
- 2526 posts since 26 Nov, 2015, from Way Downunder

How it works is like this...

1) Balance volumes and frequencies, multiband compress until just a few db of dynamic range and then multiband brickwall limit the hell out of it.

2) There is a fair amount of distortion in those tracks - you're just not hearing it.

If there's one EDM genre that would deeply benefit from increased dynamic range it's this kind of progressive trance. When the breakdown, buildup and crescendo are all the same volume, that's a wasted opportunity to add impact and emotion.

1) Balance volumes and frequencies, multiband compress until just a few db of dynamic range and then multiband brickwall limit the hell out of it.

2) There is a fair amount of distortion in those tracks - you're just not hearing it.

If there's one EDM genre that would deeply benefit from increased dynamic range it's this kind of progressive trance. When the breakdown, buildup and crescendo are all the same volume, that's a wasted opportunity to add impact and emotion.

- KVRian
- 693 posts since 6 Jan, 2017, from Outer Space

Advice on loudness war is not a good advice. We need more dynamics not less. The video about the creator lacks music which would reflect the nature of nature...

Limiters are way overrated, they are rarely needed, unless you master for the car radio, but the radio station will do that for you anyway...

The first track has a unnatural kick without the slightest variation, if that is loud and short, you will have 0dB throughout the track, and can still place stuff "behind" it without adding to the overall level, especially if you sidechain the the kick to the other tracks...

Better than being loud all the time though is to have dynamics with some sort of fibonacci rules...

Limiters are way overrated, they are rarely needed, unless you master for the car radio, but the radio station will do that for you anyway...

The first track has a unnatural kick without the slightest variation, if that is loud and short, you will have 0dB throughout the track, and can still place stuff "behind" it without adding to the overall level, especially if you sidechain the the kick to the other tracks...

Better than being loud all the time though is to have dynamics with some sort of fibonacci rules...

- KVRAF
- 10507 posts since 12 Jun, 2004, from Alberto Balsam

Copypasting my last post in a thread we just had on this.

Clean up your mix as much as you can. Highpass every track as much as you can, scoop out as many unneeded frequencies as much as you can.

Compress each track to the point of detriment and then move upwards doing the same to the buses.

Side-chaining helps a lot with loudness as well. You'll notice when listening to pop music, you really only hear one or two things at any given instance, because everything is moving out of the way of each other.

Keep in mind almost none of this is good mixing advice, just good loudness advice. But that's pop music for you.

Clean up your mix as much as you can. Highpass every track as much as you can, scoop out as many unneeded frequencies as much as you can.

Compress each track to the point of detriment and then move upwards doing the same to the buses.

Side-chaining helps a lot with loudness as well. You'll notice when listening to pop music, you really only hear one or two things at any given instance, because everything is moving out of the way of each other.

Keep in mind almost none of this is good mixing advice, just good loudness advice. But that's pop music for you.

- KVRAF
- 11855 posts since 7 Dec, 2004

Note that mixing (addition) can also be negative (subtraction).

The result of A - B = depends upon the inverse of correlation, but for a correlation of zero = 1.

A great example of this is when you subtract two pulse (square) waves. In-phase with matching frequency they'll be 100% correlated and the result will be zero of course.

Where they're not fully correlated however such as when slightly out-of-phase the result can be of equal peak amplitude and only differing amplitude of individual harmonic partials.

Due to this fact you can get some very interesting results using subtraction and phase shifts.

Here is a great example of something like this:

The second waveform has the same audible effect as two detuned "up saw" ramp waveforms even though it has a peak amplitude of exactly 1.

The result of A - B = depends upon the inverse of correlation, but for a correlation of zero = 1.

A great example of this is when you subtract two pulse (square) waves. In-phase with matching frequency they'll be 100% correlated and the result will be zero of course.

Where they're not fully correlated however such as when slightly out-of-phase the result can be of equal peak amplitude and only differing amplitude of individual harmonic partials.

Due to this fact you can get some very interesting results using subtraction and phase shifts.

Here is a great example of something like this:

The second waveform has the same audible effect as two detuned "up saw" ramp waveforms even though it has a peak amplitude of exactly 1.

Free plug-ins for Windows, MacOS and Linux. Xhip Synthesizer v8.0 and Xhip Effects Bundle v6.7.

- KVRer
- 21 posts since 30 Aug, 2005, from Finland, Tuusula

MogwaiBoy wrote:How it works is like this...

1) Balance volumes and frequencies, multiband compress until just a few db of dynamic range and then multiband brickwall limit the hell out of it.

2) There is a fair amount of distortion in those tracks - you're just not hearing it.

1. I have and idea that this could be the thing, but then again I have tested it and I can't make it work I think I maybe know why, but I would have to remaster the track to make this work. Yeah there is distortion offcourse and we can't hear it. There is always distortion when limiting the overall volume of a track, but the thing here was how to do it in the way that it's not audible distortion and still get the track peak to 0db from the start

- KVRian
- 1073 posts since 25 Aug, 2012

Formosus wrote:MogwaiBoy wrote:How it works is like this...

1) Balance volumes and frequencies, multiband compress until just a few db of dynamic range and then multiband brickwall limit the hell out of it.

2) There is a fair amount of distortion in those tracks - you're just not hearing it.

1. I have and idea that this could be the thing, but then again I have tested it and I can't make it work I think I maybe know why, but I would have to remaster the track to make this work. Yeah there is distortion offcourse and we can't hear it. There is always distortion when limiting the overall volume of a track, but the thing here was how to do it in the way that it's not audible distortion and still get the track peak to 0db from the start

It's perceived loudness you need to explore. RMS meter is the most important one for dat. You gotta get it high as possible. This video will explain...

https://www.youtube.com/watch?v=5YOrH_nNSxs

But before you even get to mastering you gotsta make Z mix levels so the frequencies are balanced. It's jigsaw puzzle and each piece is a frequency band, they gotsta lock in evenly. That's the name of the game. You dig?

- KVRian
- 883 posts since 30 May, 2003, from Paris

Kinh wrote:It's perceived loudness you need to explore. RMS meter is the most important one for dat. You gotta get it high as possible.

This place never ceases to fry my brain, it's like the blind leading the blind...