|
Help Me Understand The Frequency Analyzer
|
View this Thread in Original format
| cenik |
So yesterday I went over to a buddy's house and he showed me a few Waves plugins and stressed over and over again the importance of analyzing the frequencies being emitted by your sounds and using an equalizer to adjust them accordingly. I thought I understood the basics of what he was saying but now that I'm playing around in Live I realize that I'm rather confused. Here's a screen-shot of a frequency analyzer (on the left) and an equalizer (on the right) for a bass-line that I have soloed in Live:
Initially, I thought the idea was to smooth out the peaks and valleys in the frequencies by using an equalizer (like the one on the right in the picture) to adjust the levels of different frequencies accordingly. But now I'm not so sure this is the point of analyzing the frequencies. Would somebody please be kind enough to take a few minutes to explain to me what the goal/aim/purpose is re: analyzing frequencies and using equalizers to adjust them? I'd appreciate any help on the topic (especially since I'm under the impression that equalization is a rather key part of production...bare with me, I'm a newb). Thanks :) |
|
|
| Subtle |
| In my experience, using those frequency analyzers gives me roughly the same results every time. |
|
|
| DigiNut |
No, no, no, no, no.
Don't try to EQ a track with your eyes instead of your ears, and definitely never ever try to use EQ to kill the peaks. The peaks are the sound! You try to flatten those and you'll basically end up with mud.
The point of EQ is to get all of your various track elements to sit in the mix together and not drown each other out. For example, you might reduce the lows in one of your lead synths that conflict with the bassline, which would give you some more headroom and allow you to raise the overall volume of your mix. Or the opposite - EQ out the highs from your kick and/or bassline so you can play your melodic instruments quietly but still keep them audible.
Generally the only time I resort to using a frequency analyzer is when the mix has weird transients (like suddenly clips to +2 dB when it normally stays below 0) and I can't figure out where it's coming from. Throwing up a spectrum analyzer can help reveal where the transient is, after which you can try to EQ it out on the guilty track or just use automation/compression to squelch it.
An analyzer can also help if you're trying to bring out a sound. You find where the higher peaks are and raise those frequencies slightly (very slightly, rarely more than +1 dB or so), which help the single instrument cut through the mix without killing the mix or making it clip.
Edit: In thinking about this, the main point to consider is that preset patches and even most sounds producers synthesize themselves are initially put together to sound good solo. However, in a mix, only a tiny fraction of what's in that sound is truly interesting, and the rest is drowned out by the other instruments. If it's going to be drowned out anyway, why not eliminate it entirely to get a cleaner and louder mix?
But to be honest, when you say you're not really sure what the point of a spectrum analyzer is, that's not such a bad thing. Most of the time it's better to make these adjustments by ear. You only need a spectrum analyzer when you're sure that everything sounds fine but the mix isn't behaving quite the way you'd like it to (generally it's a clipping problem but might also be phasing, which PAZ also shows if you bust out the "full" one). |
|
|
| cenik |
| quote: | Originally posted by DigiNut
No, no, no, no, no.
Don't try to EQ a track with your eyes instead of your ears, and definitely never ever try to use EQ to kill the peaks. The peaks are the sound! You try to flatten those and you'll basically end up with mud.
The point of EQ is to get all of your various track elements to sit in the mix together and not drown each other out. For example, you might reduce the lows in one of your lead synths that conflict with the bassline, which would give you some more headroom and allow you to raise the overall volume of your mix. Or the opposite - EQ out the highs from your kick and/or bassline so you can play your melodic instruments quietly but still keep them audible.
Generally the only time I resort to using a frequency analyzer is when the mix has weird transients (like suddenly clips to +2 dB when it normally stays below 0) and I can't figure out where it's coming from. Throwing up a spectrum analyzer can help reveal where the transient is, after which you can try to EQ it out on the guilty track or just use automation/compression to squelch it.
An analyzer can also help if you're trying to bring out a sound. You find where the higher peaks are and raise those frequencies slightly (very slightly, rarely more than +1 dB or so), which help the single instrument cut through the mix without killing the mix or making it clip.
But to be honest, when you say you're not really sure what the point of a spectrum analyzer is, that's not such a bad thing. Most of the time it's better to make these adjustments by ear. You only need a spectrum analyzer when you're sure that everything sounds fine but the mix isn't behaving quite the way you'd like it to (generally it's a clipping problem but might also be phasing, which PAZ also shows if you bust out the "full" one). |
Thanks Digi for the truly helpful response. |
|
|
| echosystm |
analysers aren't of massive use for equing, like digi said. they can help, if your ears are failing you a bit though. you need a very high FFT in order for it to be accurate. i use voxengo glisseq alot for this reason. with this plugin, you can overlay the analysis of "sound a" (gliss eq instance 1) onto another instance of the plugin (instance 2). this is usually pretty good at showing where big collisions are. also, they're good for cleaning up sounds.
the waves paz analyzer does have a great stereo scope. this is the one analysis tool i think is almost mandatory for me - avoid anti-phase as much as posssible. you usually cant hear it unless the song is going extremely far into anti-phase, so this tool is very useful. |
|
|
| kitphillips |
| Its often better to emphasise the peaks as this is where the richest sound is. I find analysers useful only for checking if your clipping or not... My two cents. |
|
|
| Theran |
I must say I don't agree with Digi.
I always use a frequency Analyzer (FA), first of all, because you can't hear every frequency of every sound, with a FA you can check out which synths is on which frequency and can cause problems, problems that the ear doesn't always percieve. In that way you can avoid conflicts by looking at the FA.
I do agree on that you must not EQ based on what you see.
And BTW, the Analyzer from Waves really sucks. I found out that it is not accurate on analyzing the frequencies, but does give you a good view of your stereo spectrum.
For analysis I always use the Voxengo Span (link), it's free and IMHO one of the best analyzers on the market. |
|
|
| Pjotr G |
you said you're playing with a bass sound. You can't boost what isn't there; if the sound just doesn't have any high end, there is no point in trying to eq that in.
ears first! |
|
|
| Magnus |
| As Digi and others have said, trust your ears. I've learned to do that and have far better success than reading some graphs. |
|
|
| Ry Thomas |
| Out of curiosity, can anyone shed any light on how the Pinguin analysers work, cant seem to get any life out of the demo |
|
|
| zodiac9 |
| My buddy, who has become quite adept at and knowledgeable about mastering, finds the waves frequency analyzer most important at the mastering stage. He was mastering one of my tracks a while back, and pointed to some high frequencies on the analyzer that needed to be dealt with. He said those frequencies would "shred" club speakers. Well anyway, I didn't understand all of what he said, but made me see the importance of frequency analyzers. When EQing individual instruments in a track, I just use my ears. Like everyone is saying here, that is best. |
|
|
| Derivative |
| quote: | Originally posted by echosystm
analysers aren't of massive use for equing, like digi said. they can help, if your ears are failing you a bit though. you need a very high FFT in order for it to be accurate. i use voxengo glisseq alot for this reason. with this plugin, you can overlay the analysis of "sound a" (gliss eq instance 1) onto another instance of the plugin (instance 2). this is usually pretty good at showing where big collisions are. also, they're good for cleaning up sounds.
the waves paz analyzer does have a great stereo scope. this is the one analysis tool i think is almost mandatory for me - avoid anti-phase as much as posssible. you usually cant hear it unless the song is going extremely far into anti-phase, so this tool is very useful. |
They most certainly are useful for EQing.
Behold a 260hz square wave:
You can tell alot of information about this sound just by looking at it, if you know what to look for. You know it is a harmonic oscillator because of the regular spacing between harmonics and the diminishing amplitude of each one the higher up the scale you go. Since this is a constant sound (100% sustain) we can leave it to peak hold forever since the waveform is simple, digital and doesn't morph or change period or anything.
I'll need to explain a few of the buttons first.
Blk = FFT block size. Large block size = better low frequency resolution but poorer high frequency response. This should corresponds with LF rez on your spectrum analyser.
Spd = Update rate. i.e. how quickly the analyser updates the spectrum.
Slope = Bias towards high frequency. Higher frequency sound is often weaker than lower frequency sound in most music. You can 'bias' the slope to compensate for this so you can see the treble end better. If you run pink noise through the analyser it will slope gradually downwards by default. If you bias it 3dB towards HF it will appear flat (which is what you would want to do if you were trying to tune a room or something).
Weight = Add weighting function to the spectrum but I have no idea what this does and never use it. Its probably useful so I should read up about it and report back here.
Also you might look at the grid and wonder why the spacings are not constant. This is because both y and x axis scales are logarithmic.
On the x axis the scale actually starts at 20hz. Then you have 30hz before you see 50hz marked on the scale. Then you have 60, 70, 80 90 and 100hz (marked). After this you have 200hz (marked). Then 300 and 400hz before the next marker - 500hz. This scale makes it really easy to measure harmonics which are always multiples of a fundamental frequency.
The above image is the same square wave with no bias towards higher frequency.
I have tuned the filters roughly to the first 5 harmonics so you can see why this is a square wave (without even hearing it). First you know that this is in key of C because the fundamental is roughly 260hz .
Now you can tell its a square wave because every even harmonic is missing (i.e. 520hz, 1040hz etc.). There are only odd harmonics (i.e. 780hz, 1300hz etc.)
On your spectrum analyser you could tell the peak amplitude of the square wave because the fundamental frequency of any harmonic oscillator is always the highest amplitude and all the harmonics are measured in relation to it. I cant do it with my one because it is in fact a paragraphic EQ with a spectrum overlay. However, I could load up Voxengo SPAN right now and find that information (SPAN = free spectrum analyser).
The only time the fundamental is not the highest amplitude is when you get additive phasing or when you use resonance to raise the amplitude of the signal around the filter's cutoff. You can tell when this is occuring by watching a spectrum analyser with a fast update rate.
If you see harmonics that are out of place (i.e. dramatically taller than lower harmonics) then you know that narrow band frequencies are overlapping and you know precisely the point this is occuring at. You don't have to notch EQ one of them out (though that is an option). You could instead add a phase difference to prevent additive phasing in that band (but this will probably create periodic destructive and additive phasing in other bands if you arent careful. You will be able to see this 'comb filter' happening on the spectrum if it is properly set up). You could use a frequency dependant compressor to apply gain reduction to one or both harmonic peaks by plugging a band pass filter into a compressor's sidechain, tuning the cutoff of the filter to the harmonic you want to kill and apply as much gain reduction as you need to on the compressor.
If a spectrum analyser appears gibberish it is usually because you haven't got it configured properly for what you need to use it for. For low frequency sounds that you want to study intently you need to increase the FFT blocksize alot so you can actually see it properly. But if you need to look closely at high frequency sound you need to lower the FFT blocksize again because time coherance will start to go screwy the higher up the spectrum you go.
Also, you will not get much benefit from running a spectrum of an entire mix. The spectrum will just appear to be a block with a spikey roof and you won't be able to make out individual harmonics or individual instruments. I find analysers work best at the end of every channel so you can look at each sound in isolation and then compare it to the spectrums from other channels to see where problem frequency ranges are. Or better yet by using Voxengo GlissEQ which lets me overlay 4 different spectrums from 4 different mixer channels onto the one EQ. Its better on CPU load this way too.
I have problems distinguishing by ear whats going on in the upper mids on my Dyn BM5as and my mixes always turn out mid range heavy on my headphones. I have come to rely on GlissEQ and SPAN as guides that help me keep on track. You should still use your ears to make judgements since if it sounds , it sounds , period.
I consider these as important as meters. You could mix without them but why would you? Its just more feedback (even if it is visual instead of auditory). You don't have to use it if you don't want to but its there if you need it.
The problems only occur when you use the analyser incorrectly. For example, bias the spectrum like 9dB towards treble, look at the spectrum and think 'whoa, theres way too much treble. I need a low pass filter.'
By this point your ears should be telling you that there really isn't that much treble and your brain should be telling you to turn off spectrum bias before you do something dumb. And you should listen to them. No sir. The problems only occur when you don't know what you are doing and you trust your eyes over what you hear. Remember, what you see (on a spectrum) and what you hear (from your speakers) should correspond. Use both in tandem or just your ears if you trust them enough. I don't trust my ears or my monitors so heh.
Other mistakes people make is to set the FFT block size as high as it will go thinking that this offers the best resolution all the time. You will find that you can pick out alot more detail on low frequency sounds but the treble will fall off the radar sooner and be really innacurate. So don't EQ treble with a massive FFT block size without listening as well. For that matter never EQ a sound using only the image and not your ears. Never make any mixing decision using only your eyes and not your ears. If you are going to use things like meters and spectrums you need both your eyes and ears and you need to strongly correlate the image to the sound. It takes getting used to but its worth it once you get it down. |
|
|
|
|
