 |
|  basic steps of a proper mixdown? [MUSIC] | |
|
 |
Atlantis
  Member
Topics: 84
Replies: 3227
Registered: 14.Jan.03 |
 |
Angel in Ashes wrote on 09 Mar. (7:24) :
@disaster: Search the CTG Forums, there is a Mixing Tutorial written by Atlantis, should be useful |
Is there? Even so, I've written a lot of unperceptive stuff over the years, and I really don't recommend you trace back my learning curve. 
a98 wrote on 09 Mar. (9:05) :
the only nessesary mastering you have to do is "normalization" you can find that one on pretty much all audio editors. it will raise up the volume of the song, so that your highest peak is in -0.0db
if you want it even louder, then use light limiter or compressor settings, even harder use harder settings. |
Why would you even want to normalise if you're going to limit afterwards? And raising to 0.0 dB is also not a good idea as dithering can push the sample value beyond this, not to mention that some converters (so I've read) have trouble with 0.0 dB samples.
Alphadelta wrote on 09 Mar. (10:37) :
Anssi made some great points, although I personally never 'normalise' any of my tracks - I read somewhere aaaaages ago that it shouldn't be done - I personally just compress/maximise to raise the overall level.. can someone confirm this?? 
|
I answered this in another topic not long ago, so here it is again:
Normalisation causes quantisation distortion, because the original 16 bit samples (provided your file is originally stored in 16 bits) are altered in gain (normalisation) by means of a multiplication process, which increases the wordlenghts of those samples beyond 16 bits. But then what happens as you attempt to resave them (i.e. the file) in 16 bit? The lower bits are rounded off (or truncated, not sure on that one), causing the LSB (least significant bit) to fluctuate between on and off, resulting in low-level distortion. This degrades the signal and introduces artefacts common to digital audio.
The trick is to do all processing at the highest bit depth possible, only going down to 16 bits by means of dithering at the end.
disaster wrote on 09 Mar. (11:34) :
any ideas which plugin is really good for cutting off frequencies?
i think anykind of multiband eq should do the trick.
but how do i know which frequencies are not in the valid range lets say a leadsynth and a drumloop (comming in trough Reason [redrum]) ?
where do the drums frequencies start and where do they end?
|
Any EQ plugin will do, although you'll find different plugins have different sounding characteristics, suited to different applications. And there isn't actually a thing called a "multiband EQ", although you're probably referring to a parametric EQ (full control over a band's frequency, gain, and Q), or a paragraphic EQ (the same, but with a visual graph).
As for your other question, I'd say you just need to listen, but I know how hard that can be. In the end though, it's all in the years of hard work you put into it. No books will come close to your own experiences and the mistakes you make. Spectrum analysers can be a great help though.
Vegaro wrote on 09 Mar. (12:45) :
Is that so? I can't see how normalizing could be harmful but if you find out please let me know. |
It's not harmful if you keep the bit depth higher than 16 bits during processing.
a98 wrote on 09 Mar. (12:46) :
the theories about normalization harming the the audio is bullshit, atleast in range people could actually realize it. expecially if you compress/limit after that. basicly what normalization does is raises up the volume nothing else. |
Well, that may be true. But normalisation still causes quantisation distortion if you're limiting yourself to 16 bit processing, which progressively degrades the sound, making your reverb tails and other subtle details sound grainy.
i've always done all my mastering to normalized file and more experienced musicians have suggested me to do that also.
the result of normalizing and then mastering, and just mastering a silent audio are different.
basicly if you don't normalize you have to
a) use harder limiter/compessor settings (less dynamics)
b) raise the volume after you have compressed... from which both are not good things to do.
unless you want your master to be quiet, but isn't the main point of mastering that you get your master louder so it doesn't sound too quiet compared to other tracks.
making the master sound a lot better is usually something amateurs and even professionals can hardly ever do, unless the mixing is really crappy in the first place.
i'm not familier with maximise plugins? but i assume what they do is a combination of normalization and limiter? |
Again, why would you even want to normalise? It doesn't have any benefit if you're limiting afterwards. Sure you need to set the limiter threshold lower if the file isn't normalised to 0.0 dB, but does that really matter?
I'm not sure there is a textbook definition on maximisers, but what you say is probably right. At least the Waves L range automatically applies makeup gain, which you can scale using the 'ceiling' control.
a98 wrote on 09 Mar. (15:59) :
actually you should never eq your master. if your master is lacking some frequences or has too much of something, you should do that in mixdown.. find the instrument(s) which are causing them and eq them. |
I can agree with this in part, but mastering still requires you to approach the mix from a different angle, and preferably one coming from a different person. Mixing is much more concerned with the individual instruments, where mastering steps in to treat the resulting sound as a whole. No one is perfect enough to EQ a mix in such a sense that no wideband adjustments to the tonal balance are needed afterwards, and it can certainly bring in character to the sound too.
|
nothing under -0.0db clips.. and limiters never make the master clip, the most hardest limiter ever would theoreticly make the audio look like rectangle.. although most of the sounds probably twist somehow and may get distorted. |
There's still dithering though...
Atlantis [Atlantean Records]
multiband professor/eq professor/Nur-Ab-Sal/Generally, I Agree with Cooth
http://www.myspace.com/atlanteanrecords |
|
a98
 Artist
Topics: 32
Replies: 2725
Registered: 26.Oct.03 |
|
Atlantis wrote on 09 Mar. (19:43) :
why would you even want to normalise? It doesn't have any benefit if you're limiting afterwards. Sure you need to set the limiter threshold lower if the file isn't normalised to 0.0 dB, but does that really matter?
|
well i maybe wrong, but isn't the way limiter works so that it doesn't do anything (or not much) to the highest peak, but raises the volume of the lower ones based on the settings.
so if you have a real quiet mixdown (like i usually have), and you limit it with some light settings (so you don't loose your dynamics completely) the result is way more silent master than what you would have gotten if the source audio was already as loud as it can be without clipping.. ?
or can you explain to me how limiter actually works?
Normalisation causes quantisation distortion, because the original 16 bit samples (provided your file is originally stored in 16 bits) are altered in gain (normalisation) by means of a multiplication process, which increases the wordlenghts of those samples beyond 16 bits. But then what happens as you attempt to resave them (i.e. the file) in 16 bit? The lower bits are rounded off (or truncated, not sure on that one), causing the LSB (least significant bit) to fluctuate between on and off, resulting in low-level distortion. This degrades the signal and introduces artefacts common to digital audio.
The trick is to do all processing at the highest bit depth possible, only going down to 16 bits by means of dithering at the end.
|
ok i'm sure you are right since you are an expert on this field, those term don't say anything to me, neither i understood anything or not much of the whole post.
i think your perspective to mixing and mastering hasn't got much do do with music anymore. kinda like a person who tunes up a car as much as possible but never drives with it..
i wanna know the difference of result (explained for dummies way) you get:
- with normalisation and with simple raising up the volume (gain) (what's the difference)
- with normalisation and turning volume up from your stereo..
i consider myself to have pretty good hearing when it comes to music, but i seriously can't hear even a bit difference of audio that is normalized and audio that you have just boosted the volume up to same point (weather it's with your stereos, playback equipment, or volume increase in audio editing program)
so that's why i refuse to believe there is actually musical difference with normalisation 
creators of audio programs like sound forge, wavelab etc are one of the most experienced audioengineers there are, so if normalisation would cause negative things they wouldn't put that to their programs.
a98.subtraxx.com (2418 hits) |
|
LoneStar
Member
Topics: 3
Replies: 154
Registered: 03.Jun.04 |
a98 wrote on 09 Mar. (20:29) :
i wanna know the difference of result (explained for dummies way) you get:
- with normalisation and with simple raising up the volume (gain) (what's the difference)
a98.subtraxx.com (2418 hits) |
There is no difference.
Normalisation = Gain to 100%, with gain factor calculated by maximum (peak) level in the audio file.
erm... er... tech n00b. |
|
Atlantis
Member
Topics: 84
Replies: 3227
Registered: 14.Jan.03 |
|
a98 wrote on 09 Mar. (20:29) :
well i maybe wrong, but isn't the way limiter works so that it doesn't do anything (or not much) to the highest peak, but raises the volume of the lower ones based on the settings.
so if you have a real quiet mixdown (like i usually have), and you limit it with some light settings (so you don't loose your dynamics completely) the result is way more silent master than what you would have gotten if the source audio was already as loud as it can be without clipping.. ?
or can you explain to me how limiter actually works? |
This is in essence correct, but I think it's easier if you look at it the other way. A limiter works on values higher than the threshold, and 'limits', or reduces them down to the threshold. Exactly the way a compressor works, but with the ratio set to infinite, and the attack time set to instant. At the same time, any low sample values are brought up, but only if you normalise the limited file afterwards (as is automatically done in the Waves L range). In this sense, there is no need to normalise the file first, since you can just lower the threshold based on the input sound to achieve the desired amount of gain reduction.
|
i think your perspective to mixing and mastering hasn't got much do do with music anymore. kinda like a person who tunes up a car as much as possible but never drives with it.. |
There might be some truth in that actually. I just feel I need to know exactly what's going on before I can do something. Maybe not the most creative way of doing things, but then I'm just an engineer.
i wanna know the difference of result (explained for dummies way) you get:
- with normalisation and with simple raising up the volume (gain) (what's the difference) |
Both normalisation and gain are exactly the same thing, since they both involve a multiplication process, the former often only taking the level right up to maximum.
If a 16 bit sample value is 12,345, and you multiply this by 2.5, the resulting sample value is 30,862.5. But of course there is no decimal place available in 16 bits, so the 0.5 is either truncated or rounded (i.e. influencing the LSB), quantising the sample value from the higher bit depth (processing, i.e. internal mixing is done at a much higher resolution, to keep track of these decimal places) into a lower bit depth (the 16 bit result), distorting the waveform from its original shape.
|
- with normalisation and turning volume up from your stereo.. |
To be honest, this is a bit of a grey area for me still. The thing, though, is that once the audio reaches the amplifier, it is already in its electrical form rather than being stored in discrete bits, so maybe that allows for more intermediate values without affecting the dither noise? I'm really not sure, just as I'm trying to think of a better way of explaining all this, but in a way I feel I'm still far too incompetent. 
i consider myself to have pretty good hearing when it comes to music, but i seriously can't hear even a bit difference of audio that is normalized and audio that you have just boosted the volume up to same point (weather it's with your stereos, playback equipment, or volume increase in audio editing program)
so that's why i refuse to believe there is actually musical difference with normalisation
creators of audio programs like sound forge, wavelab etc are one of the most experienced audioengineers there are, so if normalisation would cause negative things they wouldn't put that to their programs. |
Maybe I should put some examples together. I'll see if I can find some suitable material.
And, sure, normalisation isn't a negative thing per se. I do use it sometimes, but I always increase the bit depth to 64 bit float before doing so, and dither afterwards (if needed) before saving the file in 16 bits.
Atlantis [Atlantean Records]
multiband professor/eq professor/Nur-Ab-Sal/Generally, I Agree with Cooth
http://www.myspace.com/atlanteanrecords |
|
a98
Artist
Topics: 32
Replies: 2725
Registered: 26.Oct.03 |
well one more question..
you say waves l-series do a normalisation afterwards, and that if you don't normalise you put extra gain in your limiters/compressors.. and if gain is same as normalisation, then why is using normalisation bad, but using compressors and limiters not?
or is it different to use normalisation after limiting than before?
oh and finally, i recommend everyone does everything based on their own knowledge and ears. i don't think it's a good idea to do what someone else says you should do (even if he's a professional) if you have no idea what you are doing and why..
a98.subtraxx.com (2418 hits) |
|
Atlantis
Member
Topics: 84
Replies: 3227
Registered: 14.Jan.03 |
|
a98 wrote on 10 Mar. (5:53) :
well one more question..
you say waves l-series do a normalisation afterwards, and that if you don't normalise you put extra gain in your limiters/compressors.. and if gain is same as normalisation, then why is using normalisation bad, but using compressors and limiters not?
or is it different to use normalisation after limiting than before? |
Yeah, I guess I hadn't explained that clearly enough. Normalisation is bad in a sense, but, when using and maintaining a high enough bit depth (so the rounding errors move into the inaudible range), normalisation is just the same as any other digital audio process, be it EQ, compression, reverb, etc., and, yes, the makeup gain on a limiter is just the same as any other gain change, which is essentially what every calculation is.
So, provided you keep the bit depth high during processing (as is in most cases done internally anyway), you can safely normalise before limiting if you want to, the only thing being that it's still a redundant process you don't necessarily need to make.
|
oh and finally, i recommend everyone does everything based on their own knowledge and ears. i don't think it's a good idea to do what someone else says you should do (even if he's a professional) if you have no idea what you are doing and why.. |
Sure, I was going to write that in my first post actually, and say that I respect your way of doing things if it works for you, but kind of forgot about it.
There are few rights and wrongs when it comes to mixing, but, if anything, you definitely don't want to take a 16 bit file, normalise it, and save it as a 16 bit file again.
Atlantis [Atlantean Records]
multiband professor/eq professor/Nur-Ab-Sal/Generally, I Agree with Cooth
http://www.myspace.com/atlanteanrecords |
|
mishmash
Member
Topics: 5
Replies: 30
Registered: 31.May.05 |
|
Atlantis wrote on 10 Mar. (6:34) :
a98 wrote on 10 Mar. (5:53) :
well one more question..
you say waves l-series do a normalisation afterwards, and that if you don't normalise you put extra gain in your limiters/compressors.. and if gain is same as normalisation, then why is using normalisation bad, but using compressors and limiters not?
or is it different to use normalisation after limiting than before? |
Yeah, I guess I hadn't explained that clearly enough. Normalisation is bad in a sense, but, when using and maintaining a high enough bit depth (so the rounding errors move into the inaudible range), normalisation is just the same as any other digital audio process, be it EQ, compression, reverb, etc., and, yes, the makeup gain on a limiter is just the same as any other gain change, which is essentially what every calculation is.
So, provided you keep the bit depth high during processing (as is in most cases done internally anyway), you can safely normalise before limiting if you want to, the only thing being that it's still a redundant process you don't necessarily need to make.
|
oh and finally, i recommend everyone does everything based on their own knowledge and ears. i don't think it's a good idea to do what someone else says you should do (even if he's a professional) if you have no idea what you are doing and why.. |
Sure, I was going to write that in my first post actually, and say that I respect your way of doing things if it works for you, but kind of forgot about it.
There are few rights and wrongs when it comes to mixing, but, if anything, you definitely don't want to take a 16 bit file, normalise it, and save it as a 16 bit file again.
Atlantis [Atlantean Records]
multiband professor/eq professor/Nur-Ab-Sal/Generally, I Agree with Cooth
http://www.myspace.com/atlanteanrecords |
'There are few rights and wrongs when it comes to mixing, but, if anything, you definitely don't want to take a 16 bit file, normalise it, and save it as a 16 bit file again....'
Sh*t, now they tell me...
|
|
Atlantis
Member
Topics: 84
Replies: 3227
Registered: 14.Jan.03 |
mishmash wrote on 11 Mar. (21:23) :
'There are few rights and wrongs when it comes to mixing, but, if anything, you definitely don't want to take a 16 bit file, normalise it, and save it as a 16 bit file again....'
Sh*t, now they tell me... |
It's the same with any digital process though. If you apply EQ or reverb to a file, you're increasing the wordlength of the samples. And if you then quantise those samples back into the original 16 bit range (i.e. saving the file in a lower bit depth from processing), you're distorting the waveform, introducing a grainy character that degrades the width and depth of the music.
Atlantis [Atlantean Records]
multiband professor/eq professor/Nur-Ab-Sal/Generally, I Agree with Cooth
http://www.myspace.com/atlanteanrecords |
|
Analysis
Artist
Topics: 72
Replies: 6282
Registered: 16.Mar.03 |
|
Atlantis wrote on 11 Mar. (21:48) :
mishmash wrote on 11 Mar. (21:23) :
'There are few rights and wrongs when it comes to mixing, but, if anything, you definitely don't want to take a 16 bit file, normalise it, and save it as a 16 bit file again....'
Sh*t, now they tell me... |
It's the same with any digital process though. If you apply EQ or reverb to a file, you're increasing the wordlength of the samples. And if you then quantise those samples back into the original 16 bit range (i.e. saving the file in a lower bit depth from processing), you're distorting the waveform, introducing a grainy character that degrades the width and depth of the music.
|
Seriously? Sh*t, I didnt release this at all. Does this also account for effects inside a program? Coz I've got no clue how to increase bitrate after each effect that I plugin, and even when it doesn't.. How could I increase the bitrate in soundforge after each effect that I use?
 Or did I misunderstood anyone here?
Don't take yourself to serious |
|
Atlantis
Member
Topics: 84
Replies: 3227
Registered: 14.Jan.03 |
|
Analysis wrote on 12 Mar. (5:03) :
Atlantis wrote on 11 Mar. (21:48) :
mishmash wrote on 11 Mar. (21:23) :
'There are few rights and wrongs when it comes to mixing, but, if anything, you definitely don't want to take a 16 bit file, normalise it, and save it as a 16 bit file again....'
Sh*t, now they tell me... |
It's the same with any digital process though. If you apply EQ or reverb to a file, you're increasing the wordlength of the samples. And if you then quantise those samples back into the original 16 bit range (i.e. saving the file in a lower bit depth from processing), you're distorting the waveform, introducing a grainy character that degrades the width and depth of the music.
|
Seriously? Sh*t, I didnt release this at all. Does this also account for effects inside a program? Coz I've got no clue how to increase bitrate after each effect that I plugin, and even when it doesn't.. How could I increase the bitrate in soundforge after each effect that I use?
Or did I misunderstood anyone here? |
If you're applying effects inside a sequencer, the wordlengths are usually maintained in 32 bit floating point or higher, so you shouldn't have to worry about anything until it comes to the final render. If you're rendering in 16 bit, you should be dithering the file down, which is usually also done internally. If you choose to bypass the dither, you should either be using a dither plugin at the output (e.g. the Waves L3 set to dither to either 16 or 24 bit, depending on your output setting), or rendering in 32 bit floating point and dithering the file only after normalising or mastering.
If you're using Sound Forge to process a sound with effects, the internal process is maintained using 64 bit floating point, however upon saving the file in either 16 or 24 bit, you're quantising the bit depth down and introducing distortion. Unless you don't intend to process the sound further, you should be saving the file in at least 32 bit floating point, preferably 64 bit floating point. Being cynical, I still prefer to increase the bit depth to 64 bit floating point (Process->Bit-Depth Converter) before applying effects, but I don't think this should make a difference. On the other hand, if you don't intend to process the sound any further (e.g. mastering), or have no choice but to use 24 bit or 16 bit, you should be dithering as the final step, and saving your file in your chosen depth.
Atlantis [Atlantean Records]
multiband professor/eq professor/Nur-Ab-Sal/Generally, I Agree with Cooth
http://www.myspace.com/atlanteanrecords |
|
|
|
|
|
Latest Topics
![[203.109.202.28]](pics/ip.gif){kind=small}
