F
faderbug
.
Blue Bear Sound said:
i gave you an example : when an audio file is too quiet (eg - 40 db) i normalise it!! how can you not see the use of that?
i gave you an example : when an audio file is too quiet (eg - 40 db) i normalise it!! how can you not see the use of that?
E
Ed Dixon
New member
If normalizing were "useless processing", no software companies would waste their time developing normalization routines, no studios would advertise they have the capability, and no hardware firms would build it into their gear.
The fact is that normalization is NOT "useless processing". It is a process that has it advantages and is used in various parts of the recording/mastering process in virtually every studio in the world today. Trying to argue otherwise is a further waste of time, as most of us would probably prefer to use it recording in our home studios.
Ed
The fact is that normalization is NOT "useless processing". It is a process that has it advantages and is used in various parts of the recording/mastering process in virtually every studio in the world today. Trying to argue otherwise is a further waste of time, as most of us would probably prefer to use it recording in our home studios.
Ed
Blue Bear Sound
New member
If your audio file is down at -40dBFS - you've got problems anyways because the noise floor of budget gear isn't too far away from that!wetteke said:
If you normalize a file like that - you will be raising the noise floor as - to a degree that would make the audio unusable...
So again - an example of a useless process...
Blue Bear Sound
New member
Sorry Ed, you're dead wrong -- it isn't typically used at most studios at all, and virtually never (I'll say it again - NEVER) used in the mastering process by any serious facility.Ed Dixon said:
It IS typically used by novices who don't know better, though....!
Note that I say this without malice or being insulting -- it's simply a fact that the use and consequences of normalizing are counter-intuitive and as such, are widely misunderstood - particularly for those new to the field of recording.
But - I've explained the implications and given very correct reasons of why it SHOULDN'T be used.... what you do with that bit of knowledge is completely up to you.
Shailat
New member
I'm not going to waste time arguing if Normalizing is a usefull tool (but B.Bear is right).
Ron82,
To answer your question.......I have checked and checked again.....a file that has been normalized in Pro tools, is degraded compared to a file brought to the same level by other means.
Ron82,
To answer your question.......I have checked and checked again.....a file that has been normalized in Pro tools, is degraded compared to a file brought to the same level by other means.
E
Ed Dixon
New member
Blue Bear Sound said:
Sorry, but I disagree. I fully understand how normalization works and why is is needed. All one has to do is look at ads from professional studios to see references to it's use. It's been a useful process since the concept was first created.
One wants to agrue points, terms like "useless processing" don't help your case. Trying to better understand the situations where it is used effectively and when it not are better approaches.
Ed
AlChuck
Well-known member
Reading through this thread with some interest... Blue Bear, you first said, "normalizing works only on peak levels, " but then in the next sentence you said, "Normalizing is simply a gain-multiplier added to the signal that represents the difference between some maximum (0 dBFS is most common) and the highest peak of the signal. So if the signal's highest peak is -2dBFS, then a +2dB increase will be applied."
I was under the impression that normalizing simply took that difference you described (between the chosen limit "ceiling" and the peak amplitude in the data) and raised all the sample amplitudes by that value. That's what the second sentence I quotd implies, but when you say it "works only on peak levels," that sounds contradictory.
If my impression is correct, it seems like there is little processing going on other than adding amplitudes, which shouldn't mangle the data any, really. Of course, as you pointed out, it sure does make the noise level more apparent, but I can see how if your levels are good, a little normalizing might be helpful to tweak a track with just a nudge -- liek compressing it but without altering its dynamic range.
Anyway, I'm going go off and read a little and see if my idea of normalizing is off (I suspect it is...).
I was under the impression that normalizing simply took that difference you described (between the chosen limit "ceiling" and the peak amplitude in the data) and raised all the sample amplitudes by that value. That's what the second sentence I quotd implies, but when you say it "works only on peak levels," that sounds contradictory.
If my impression is correct, it seems like there is little processing going on other than adding amplitudes, which shouldn't mangle the data any, really. Of course, as you pointed out, it sure does make the noise level more apparent, but I can see how if your levels are good, a little normalizing might be helpful to tweak a track with just a nudge -- liek compressing it but without altering its dynamic range.
Anyway, I'm going go off and read a little and see if my idea of normalizing is off (I suspect it is...).
S
Slaps Forehead
New member
I'm kinda with Al (but not in a gay way 
)
If my track peaks at -5db, all I want to do is apply gain, and I know my target peak amplitude (say -2 db), how is peak norming to -2db different from adding 3db of gain?
Peak norming seems to be useful shorthand for bringing tracks in line with each other for a rough first mix, and for maxing volume without changing dynamics. Bruce, are you saying it introduces more errors than applying gain? BTW, my environment is Cubase 5 / Sound Forge, if it matters.
I've been a bit more judicious with my DSP since learning (here) about digital errors, and try to apply norming just once - and, if the whole song is tracked in Cubase, I'll often skip norming and adjust the master faders instead (targeting -1db peak, usually). But I still like norming for bringing tracks into line.
)If my track peaks at -5db, all I want to do is apply gain, and I know my target peak amplitude (say -2 db), how is peak norming to -2db different from adding 3db of gain?
Peak norming seems to be useful shorthand for bringing tracks in line with each other for a rough first mix, and for maxing volume without changing dynamics. Bruce, are you saying it introduces more errors than applying gain? BTW, my environment is Cubase 5 / Sound Forge, if it matters.
I've been a bit more judicious with my DSP since learning (here) about digital errors, and try to apply norming just once - and, if the whole song is tracked in Cubase, I'll often skip norming and adjust the master faders instead (targeting -1db peak, usually). But I still like norming for bringing tracks into line.
Blue Bear Sound
New member
It "works on peaks" meaning from the highest peak value of a waveform to the maximum (either 0dBFS or a user-defined limit).... either I said it badly or you misunderstood - I don't see where my contradiction is.AlChuck said:
It mangles the values because ANY DSP (which are mathemtics applied to the data) introduces a degree of round-off error which changes the nature of the data itself.AlChuck said:
Then on top of that, the noise floor is also increased.
Last edited:
Blue Bear Sound
New member
We'll have to agree to disagree then, Ed - because my experience is the complete opposite!Ed Dixon said:
F
Fusion2
New member
you guys don't think soft is hyped by makers and resellers? you guys sound like resellers to me...
ie: the ahem, enhancer, BBE sonic maxizer, yes, not the same thing but surely hyped up crap...
agreed, soft has come a long way but given the protools report i still say over processing is a negitive...
now it may be usefull if your using very poor recording equip, if it's at all possile to enhance it from it's original rip, it may sound better, but what does it do to a great rip from high end equip? humm, a two edged sword imho...
there's two many factior's involved in this, and hype, it will lead to flame imho...
peace...
ie: the ahem, enhancer, BBE sonic maxizer, yes, not the same thing but surely hyped up crap...
agreed, soft has come a long way but given the protools report i still say over processing is a negitive...
now it may be usefull if your using very poor recording equip, if it's at all possile to enhance it from it's original rip, it may sound better, but what does it do to a great rip from high end equip? humm, a two edged sword imho...
there's two many factior's involved in this, and hype, it will lead to flame imho...
peace...
F
faderbug
.
Blue Bear Sound said:
when i record on minidisc some of the ambient noises i get are -40db. i have very little noise on these recordings -btw, what ever gives you the idea i only use budget gear - and if it is too loud i apply some noise reduction (a useless and redicules process, i'm sure).
you must have missed the part where i wrote : 'i just LOVE processing'. for you changing the original recording is blasphemy. i don't give a shit about conserving all the original qualities of a piece of sound, i filter it until it sounds the way i want it to sound.
but even if you want to keep all the original sound qualities,
WHAT DOES IT MATTER IF YOU MAKE SOMETHING A BIT LOUDER???
Blue Bear Sound
New member
If you care about retaining sonic integrity of the signal, it matters a great deal on what DSP you apply to it..... if you don't care, then it doesn't matter one bit!
F
faderbug
.
only purists care about sonic integrity.
Blue Bear Sound
New member
Sound engineers who take their profession seriously tend to care about sonic purity at all levels of the signal chain... or at the very least, be knowledgeable enough to know when it's reasonable to ignore best practices and when it isn't.....
F
Fusion2
New member
wetteke said:
yeah, it shows in todays newer music audio... Audioslave, etc etc...
E
Ed Dixon
New member
Most folks here are looking for approaches that will yield the best audio result on their home systems. Given the equipment they have, the software available, it then boils down to finding the combination that yields the best results given the time they have to spend on the effort.
A number of steps usually make up the process. As some point one usually has 2 or more track/WAVs that are to be mixed in the final product. For most, the total is probably more like 5-16 such tracks. In a perfect world, each of these tracks would be at perfect identical levels and the mixing process would be fairly simple.
However the real world is not like that. One quite often ends up with a variety of tracks from different sources at different audio levels. Some may be clean and some might be somewhat noisy. If for example, the vocal tracks are at a much lower level than the drums and bass tracks, mixing can become somewhat harder. Thus enters the world of post processing, effects, and other similar processes.
The mixing process combines the various WAV files to produce a single WAV (usually stereo). The mixer level meters effectively tell the PC software how to combine the levels of the different WAVs present to make up the final result. If a mixer track value is other than the default value, then the samples from that WAV will be increased/decreased as they are added to the final result.
Many find that the normalization process can make this mixing step somewhat easier. Normalizing a low level vocal track will bring it’s level back up to something closer to the other existing tracks. Then when the mix down process occurs, the vocal tracks have to be modified less (if at all) to fit in with the other data already there at higher volumes.
It’s also worthwhile to point out that there are multiple approaches to WAV normalization. Various algorithms have been developed over the years and some work better than others. One has to listen to the result to decide if what they have works to their satisfaction. Some software offers flexibility on the normalization process where the user can set various parameters (peak limiting, max boost, percent target, and other values). These kind of options usually indicate a more flexible and quality normalization product.
When producing an entire CD, an additional normalization step is usually also used to make the various CD tracks play at similar volumes. If one does a search in Google, you see many hits mentioning this aspect.
Each sound engineer has to decide what is best for their effort. There is usually more than one way to “skin the cat” and each decides how best to proceed.
Some of us have probably spent some time inside real professional recording studios. While their equipment is much more expansive and their skill level higher, their process is quite similar. It’s quite fascinating to stand in a studio, play a guitar track, and then watch the engineer incorporate the result into the final mixed CD product.
Ed
A number of steps usually make up the process. As some point one usually has 2 or more track/WAVs that are to be mixed in the final product. For most, the total is probably more like 5-16 such tracks. In a perfect world, each of these tracks would be at perfect identical levels and the mixing process would be fairly simple.
However the real world is not like that. One quite often ends up with a variety of tracks from different sources at different audio levels. Some may be clean and some might be somewhat noisy. If for example, the vocal tracks are at a much lower level than the drums and bass tracks, mixing can become somewhat harder. Thus enters the world of post processing, effects, and other similar processes.
The mixing process combines the various WAV files to produce a single WAV (usually stereo). The mixer level meters effectively tell the PC software how to combine the levels of the different WAVs present to make up the final result. If a mixer track value is other than the default value, then the samples from that WAV will be increased/decreased as they are added to the final result.
Many find that the normalization process can make this mixing step somewhat easier. Normalizing a low level vocal track will bring it’s level back up to something closer to the other existing tracks. Then when the mix down process occurs, the vocal tracks have to be modified less (if at all) to fit in with the other data already there at higher volumes.
It’s also worthwhile to point out that there are multiple approaches to WAV normalization. Various algorithms have been developed over the years and some work better than others. One has to listen to the result to decide if what they have works to their satisfaction. Some software offers flexibility on the normalization process where the user can set various parameters (peak limiting, max boost, percent target, and other values). These kind of options usually indicate a more flexible and quality normalization product.
When producing an entire CD, an additional normalization step is usually also used to make the various CD tracks play at similar volumes. If one does a search in Google, you see many hits mentioning this aspect.
Each sound engineer has to decide what is best for their effort. There is usually more than one way to “skin the cat” and each decides how best to proceed.
Some of us have probably spent some time inside real professional recording studios. While their equipment is much more expansive and their skill level higher, their process is quite similar. It’s quite fascinating to stand in a studio, play a guitar track, and then watch the engineer incorporate the result into the final mixed CD product.
Ed
AlChuck
Well-known member
"Works only on peak levels" sounds to me like "has an affect only on peak levels," implying it's leaving others alone.
But the only algorithm would be add the integer x to each sample amplitude -- no real DSP math going on, hence no round-off...
Blue Bear Sound
New member
Ah, I see... no - I meant that it uses peaks...AlChuck said:
No... it's not limited to integer values at all... you can add 0.75dB of gain -- DEFINITE round-off error there!AlChuck said:
Blue Bear Sound
New member
Just because google has a high search count, doesn't make something a fact!Ed Dixon said:
And I already explained that normalization DOES NOT give you the intended results of equalizing levels between tracks....... your comment is simply wrong - period. LIMITING is what allows for level balancing across tracks - and limiting is NOT the same as normalizing.