Level Issues

mini_molko

Member
Joined
Mar 7, 2008
Basically, I'm working in Cubase 5 and getting confused about a couple of things concerning levels, decibels and limiting. Firstly, I've been having problems with getting my subs to punch through on the mix, I'd tried compressing them and then cranking up the gain through the compressor. This just resulted in maxing out my overall mix output, I then tried putting a limiter over the compressed sub and for some reason, the sub now comes through at a decent level (And pretty high on a spectrum analyser compared to the other frequencies in the track), but it isn't maxing out my overall mix output... :/

I'm not that familiar with limiting and I'm still getting my head around compression, so could someone please tell me how limiting achieves the effect of raising the volume without causing it to clip...?

Secondly, I don't really understand the concept of decibels... Is there a different system for measuring decibels when it comes to music, as I always thought 0dB was silence and the louder the sound the higher the decibel, yet a frenquency analyser seems to measure decibels in the opposite direction...

Also, to add to further confusion, if 0dB is the peak, then why does Cubase seem to clip out at about 12dB (According to a frequency analyser)?

These are probably all complete n00b questions and should be obvious to an amateur, but I guess they're things I never really picked up on when I started producing. :/
 
I hope you're not pushing any levels above 0dB... in the digital audio world everything is done in negative dB (i know confusing) but its just how it goes. Try mxing you kick and snare first... Get the kick peaking at about -6 dB and then get the snare nice and punchy with it around the same level... then mix in the sub and mid range bass so they all sound good together and everything else can sit lower dB to fill out the track... that should help you out.
 
Cheers dude, it's only just clicked that my frenquency analyser was just analysing the volume of the seprate frequency elements of the track in relation to the rest of the track and not the actual output volume of the track, which clears a few things up.
That whole decibel thing is confusing, is there a specific reason for measuring decibels in this way?
I still don't understand why putting a limiter on my sub has stopped the output of the track from clipping. I've read up a bit on limiting and understand how it reduces peaks so that lower levels can be pulled up in the mix, but the frequencies I needed to increase were the highest peaking in the first place and now they're at a substantional level without clipping. Just doesn't make sense!

---------- Post added at 23:25 ---------- Previous post was at 22:35 ----------

Ok, scrap that, I got confused - I took the limiter off and it appears as though it wasn't actually having any effect on the sub, it's still at decent level without clipping the overall output of the track with just the compressor on, so that's all good... I think.
But now I've noticed another problem; it appears as though the kick is causing the track to clip, but it's peaking at a significantly lower level than the sub according to my spectral analyser... This is even more confusing... The sub's peaking at about 3dB, yet the kicks peaking at about 12dB, I've soloed the sub and no clipping, yet I solo the kick and it clips! What the fuck's that about!?
It may be that the peak of the kick is occurring for such a short amount of time that I'm just not able to see when it peaks, is that possible?
 
It's all in the basics of digital sound. Look up a book, there are plenty that explain the very basics of it. But here's the main things:

That whole decibel thing is confusing, is there a specific reason for measuring decibels in this way?

In digital realm, you have a finite amount of possibilites of how loud a signal can be at any given time. You have a roof of how loud things can be, as you seem to understand with the 'maxing the output' thing. If you zoom in very close to a sound sample in an editor you'll see 2 things. In 16bit signal, there are some 65000 different levels of loudness (vertical place of the points) with every given sample point (the dots in the sound) in time (the horizontal axis). This is a representation of the analog world where sound is waves. So the highest place of the point is the maximum volume -0, and the lowest possible point is silence, -inf respectively. 65000 may sound like a lot but it's not. Analog world doesn't work in set points, so there will be approximation of value to the nearest point and this will cause noise.

Got a little carried away there but it's all got to do with the negative sound values so I thought I'd explain it a little.


I still don't understand why putting a limiter on my sub has stopped the output of the track from clipping. I've read up a bit on limiting and understand how it reduces peaks so that lower levels can be pulled up in the mix, but the frequencies I needed to increase were the highest peaking in the first place and now they're at a substantional level without clipping. Just doesn't make sense!

Limiters work regardless of frequencies. They get triggered by any and all frequencies, it's just a matter of how loud the signal is. I'm not sure I understand what you mean by this though. In practice, limiting and compression are the same thing. Limiter is just a compressor with 0 attack and a very high ratio.

But now I've noticed another problem; it appears as though the kick is causing the track to clip, but it's peaking at a significantly lower level than the sub according to my spectral analyser... This is even more confusing... The sub's peaking at about 3dB, yet the kicks peaking at about 12dB, I've soloed the sub and no clipping, yet I solo the kick and it clips! What the fuck's that about!?

Your analyser might be a little late or not detect maximum peaks, only average peaks. Can you set the graph to show a maximum peak values? In Voxengo span this is done by the lower right corner, labeled as "Peak hold" or something similiar.

Hmm i just read what you write after that, and I think you're on the right track.
 
Last edited:
It's all in the basics of digital sound. Look up a book, there are plenty that explain the very basics of it. But here's the main things:



In digital realm, you have a finite amount of possibilites of how loud a signal can be at any given time. You have a roof of how loud things can be, as you seem to understand with the 'maxing the output' thing. If you zoom in very close to a sound sample in an editor you'll see 2 things. In 16bit signal, there are some 65000 different levels of loudness (vertical place of the points) with every given sample point (the dots in the sound) in time (the horizontal axis). This is a representation of the analog world where sound is waves. So the highest place of the point is the maximum volume -0, and the lowest possible point is silence, -inf respectively. 65000 may sound like a lot but it's not. Analog world doesn't work in set points, so there will be approximation of value to the nearest point and this will cause noise.

Got a little carried away there but it's all got to do with the negative sound values so I thought I'd explain it a little.




Limiters work regardless of frequencies. They get triggered by any and all frequencies, it's just a matter of how loud the signal is. I'm not sure I understand what you mean by this though. In practice, limiting and compression are the same thing. Limiter is just a compressor with 0 attack and a very high ratio.



Your analyser might be a little late or not detect maximum peaks, only average peaks. Can you set the graph to show a maximum peak values? In Voxengo span this is done by the lower right corner, labeled as "Peak hold" or something similiar.

Hmm i just read what you write after that, and I think you're on the right track.

Wow, thanks, everything's starting to make a little bit more sense... I've got Voxengo span, so I'll have a go and see if I can get it to show the maximum peaks.
 
dont compress your sub to much, EQ out dodgy freqz in your sub and bass, esp between the sub and kick, and the kick and snare, the better you EQ, the less compression you will need to get it punchy, the less compression you need the fatter it'll sound

Limters can work really well for bass, but you dont need it imo for fat and punchy bass, just gentle compression and very careful EQ

adding to Kama's reply....keep all your levels below 0, including the master, the more space you give yourself in the mix, the better everythin will sound, and it'll give you lots of room for post processing like compression/multiband/limiting
 
Last edited:
I think most of what you've asked has already been covered but I think you benefit from 2 things: firstly have a look at this thread on dubstep forum. It's about proper gain structure and if you follow the advice from Macc no individual element in you mix (or the whole mix) should be in any danger of clipping. Secondly analyzers can be useful but it sounds like you're maybe relying on them a little too much, listen to what you're doing to the sound and don't rely upon an analyzer to tell you what to do.

To expand a bit on dB, dB is a relative measurement. 12dB is meaningless figure unless you compare it to something. Inside your daw it uses dBFS to measure volume - the FS stands for full scale, so dBFS is comparing volume to the highest level we can encode in digital audio - 0 dBFS. dB SPL is compared to the sound pressure level of a sound. 0.0002 dyne/cm² = 0 dB which is the threshold of human hearing.
 
i didnt read thread cos tl:dr but i handed a few tracks to a guy for mastering and he wondered how we get levels of everything exactly the same in all the tunes, he asked if we normalize all tracks or use a limiter or whatever and its none of that, its just down to hans being allergic to the vu going in the red on the channels and balancing the sounds so that certain elements always are in the front of the track. then we have tommy who eqs and compresses where necessary. the dude started talking about minus decibels and rms and shit and i have no idea what that means.
 
but... minus decibels? then you cant hear shit! and rms? rms rms rms... thats a.... its the... r is for rate obviously and m is for money usually and s is for spending so its how pricey the sound is. there. figured it out. and for the record ive dipped in the finish genepool plenty of times. aint but 40 people in finland and chances are one of them is directly related to me. so genepool schmene pool you big smelly usva.
 
0db basicly means the source signal is at full volume, this is how amps work, the source volume (essentially the Voltage) is increased/decreased, if you set the source volume to 0db, ie a Line Signal of 4.7V (max line voltage) then the Amp will output at whatever max power its ment to

If you set the source volume 50% of max, like taking say 48db off the source volume (set it to -48db), then the source input is half, and the output from the amp will be half


of course Decibels aint linear, you have to increase volume by 6db to 'double' its 'apparnet' loudness

fuk it, just dont go into the red
 
Last edited:
Back
Top Bottom