frequency splitting

  • Thread starter Thread starter dobro
  • Start date Start date
dobro

dobro

Well-known member
Frequency splitting, which I learned from Chrisharris not a week ago, is magic. I've learned all sorts of stuff about my tracks that I didn't know before, and it makes it way easier to deal with EQ now. But the tracks I'm analyzing were all done on my gear, not yours.

So, here's what I'd like to propose - a frequency splitting comparison. I've done a frequency split on a single vocal track, and I'm really surprised at the results. I'm gonna post the results in the mixing forum cuz I'd like some engineers to comment on this. So if you're interested, you can check out what I found in the mixing forum, and if you're *still* interested, maybe you could do a similar frequency split on a vocal track of your own.

I've got a hunch that I stand to learn almost as much from a comparison of my track and your track(s) as I've learned already.
 
This may sound like a assy question but what have you learned by splitting the frequencies?

Does it destroy the original track?
 
I can't get the splitter to become active. It is still inactive. What am I doing wrong?
 
Ok I got it. Kinda cool! I will try to post some examples.
 
I've learned all sorts of things so far:

- at which frequencies most of the energy of my track resides.

- the fact that even though I cut lower frequencies with EQ, they still crop up way down under 50 Hz!

- the fact that all the clarity of my tracks is up above 1K, and not at the range where I'm singing at all.

- the *sound* of each of the various frequency ranges. Like chrisharris said, it's good ear training

The next step is to see if I can make a mixdown by selecting only certain frequency bands displayed in the splitter - can I get rid of EVERYTHING south of 50 Hz, for example?
 
dobro said:
can I get rid of EVERYTHING south of 50 Hz, for example?
Click to expand...

But can't that be done with a high pass filter? And now that you know where the clarity are in your voice can't that be boosted to bring it out of the mix if need be?

BTW this is very cool.
 
Scott-

Would you mind sharing how you enabled the frequency splitter?

-Chris
 
You have to left click to select the track then hold down to select the exact area you want to split then the splitter became active... The first time I selected a little outside of the track and it didn't work. You have to be in multi-track view.
 
chris - I couldn't figure it out at first either.

You take ONE track into multitrack. Then you highlight the track by left clicking it, as Scott said. THEN you left click and drag along the track to 'double highlight' it - that's what enables the frequency splitter for the track. Then you can click on Effects and Frequency Splitter. When you get the dialog, you can choose the frequency ranges you want to use - up to eight. If you choose eight, it takes quite a while to process a three-minute track.
 
Scott, I was wondering if it's possible to make a mixdown of the frequency split page, but with only certain tracks selected. Well, you can do that. After I'd split the track into 8 bands, I made a mixdown of it, muting the lowest two bands. Magic! It dumped all the problematic lower end I was dealing with on the track. What a powerful tool. You can completely eliminate any frequency range you like. Or you can reduce any frequency range you like more precisely than with a filter, just by cutting the level on a particular range on the splitter page.

"But can't that be done with a high pass filter?"

Nope. I tried that, and found that even after I'd applied a highpass filter at 95 Hz, there was still stuff cropping in both the lowest ranges: 50-100 Hz and even 0-50 Hz! It seems that filters are kinda approximate, and that they reduce rather than eliminate frequency ranges.

"And now that you know where the clarity are in your voice can't that be boosted to bring it out of the mix if need be?"

Yes, exactly. Garbanzo!

This is a more useful tool than spectral analyzers I've seen. Not only does it let you know what's going on, it's lets you CHANGE what's going on. Damn! Rock and roll! Hoochie koo!
 
Oh My:eek:

This is marvelous!

I haven't quite figured out when it'll come in handy yet, but Im sure it must.

I wonder how much fidelity you'd lose if you split the tracks into 5 or 6 bands, and then mix them down again to a new track

I don't really know why you'd want to, but I guess if you were eliminating whole bands like Dobro, this is a quick and easy way

-Chris
 
dobro said:
Scott, I was wondering if it's possible to make a mixdown of the frequency split page, but with only certain tracks selected. Well, you can do that. After I'd split the track into 8 bands, I made a mixdown of it, muting the lowest two bands. Magic! It dumped all the problematic lower end I was dealing with on the track. What a powerful tool. You can completely eliminate any frequency range you like. Or you can reduce any frequency range you like more precisely than with a filter, just by cutting the level on a particular range on the splitter page.

"But can't that be done with a high pass filter?"

Nope. I tried that, and found that even after I'd applied a highpass filter at 95 Hz, there was still stuff cropping in both the lowest ranges: 50-100 Hz and even 0-50 Hz! It seems that filters are kinda approximate, and that they reduce rather than eliminate frequency ranges.

"And now that you know where the clarity are in your voice can't that be boosted to bring it out of the mix if need be?"

Yes, exactly. Garbanzo!

This is a more useful tool than spectral analyzers I've seen. Not only does it let you know what's going on, it's lets you CHANGE what's going on. Damn! Rock and roll! Hoochie koo!
Click to expand...

This is a very awesome tool! I did the mixdown of only the frequencies that I wanted like you mentioned and WALLA! It is great!
 
I think maybe the coolest use could be in mastering. Split it into maybe 4 bands (most mastering compressors use 3 or 4), but you want different attack and release times depending on what you're compressing, right? Well, compress the low end with short attack and long release, b/c you're not really worried about transients in the low end (so the attack can be instantaneous, if you want), but the low end can be responsible for that ugly "breathing" sound if you set the release time too quickly)... then the low mids with very short attack and very quick release - just avoid "pumping" or "breathing", then the mids with medium attack and release, and the highs with longer attack and shorter release (to get drum transients.

Try this on an old mix if you're bored...it's a way to make CEP into a multiband compressor without having to understand the CEP compressor :D (It really will "punch up" the track).


(I personally hope they never change the compressor...it rules
 
Oh, and if it's not obvious, the best way to do this is to grab a tune (commercial recording) that has the sound you're after on a particular song, split it into 3 or 4 bands and save those, then split YOUR tune into the exact same frequencies, then import the 3 or 4 bands from the commercial mix. Do an A/B of each band, ESPECIALLY the low end....

This is probably the most enlightening exercise I've ever done.
 
Chris - yeah, exactly - the quick and obvious use is at the mastering stage. The mixdown needs a bit of brightening? The mixdown's a bit too dark? Select the frequency range of your choice, sir, and boost or cut to your heart's content. It works like a multiband compressor like you said, except that in addition to compressing, in Cool 2.1 you can apply ANY effect to the range in question. Not so much multiband compressing as multiband effecting. But in addition to that, there's that ability I was raving about earlier of being able to completely AXE a frequency band from a track or a mixdown. Bye bye low end rumble. Bye bye high end hiss.

Chris, you said: "I don't really know why you'd want to, but I guess if you were eliminating whole bands like Dobro, this is a quick and easy way"

To get rid of all the noise which is below the frequency range of the song. What's the lowest note in your whole mixdown? Find it, and then create a frequency range below that with the splitter. Split the track, and then delete that range you don't want. Bingo! A quieter mixdown! A cleaner mixdown with less mud! Why? Cuz you dumped the mud!

Here's something I picked up from Craig Anderton, recording guru, today (this talks to your question about highpass filters too, Scott):

"Highpass. This passes frequencies above a particular cutoff frequency. Below it, response diminishes progressively with lower frequencies. The low-cut switches found on consoles, which generally tailor mic response, are highpass filters. This is useful because with plosive sounds (P,B), mics can produce significant low frequency energy. The more lower frequencies you can cut below the range of the material being recorded, the better."

So, you see, Craig says that highpass filters 'diminish progressively' low-end frequencies, but they don't entirely eliminate it, despite the fact that he later says the more lower frequencies you can cut the better. But with the splitter, you can cut EVERYTHING you don't want below a certain point. Screw highpass filters. Lowpass filters ditto.

But hey, everybody - I would still love to compare my results with yours. Do a splitter on a simple vocal track, preferably dry. What are your results?
 
Last edited:
One more use...Instead of using noise reduction on an entire song, try using it on ONLY the high band. For some reason, even though the digital bubbling sound SHOULD be the same regardless of whether you run noise redux on a band or an entire file, it seems less apparent to me when it's only used on the high band.

I'm done, lol.
 
"Oh, and if it's not obvious, the best way to do this is to grab a tune (commercial recording) that has the sound you're after on a particular song, split it into 3 or 4 bands and save those, then split YOUR tune into the exact same frequencies, then import the 3 or 4 bands from the commercial mix. Do an A/B of each band, ESPECIALLY the low end....

This is probably the most enlightening exercise I've ever done."

I'm beginning to become very fond of you. Caramba! You're a light year or two ahead of me. You trail stardust which I collect and use... :)
 
This is an awesome thread of useful information!!!!

Great job guys... i can't wait to use this...

But one question.. call me a noobie if you'd like but, why get rid of the low end dobro? Was the low end very dirty? Or was it unrequired for your song? Just wondering why anyone would eliminate an entire frequency range from their song...?
 
dobro said:

So, you see, Craig says that highpass filters 'diminish progressively' low-end frequencies, but they don't entirely eliminate it, despite the fact that he later says the more lower frequencies you can cut the better. But with the splitter, you can cut EVERYTHING you don't want below a certain point. Screw highpass filters. Lowpass filters ditto.
Click to expand...

Don't mean to burst your bubble, because the frequency splitter is indeed a very cool and useful tool. But how do you think CoolEdit splits the track in the first place? Answer: by using the very same high- and low-pass filters that you can use elsewhere.

Go into "Edit View" for one of the frequency bands you found, and click "Analyze....Frequency Analysis." (On any track, but doing this with a band from a white noise file would be especially revealing.) Particularly if you look at a Linear View (not Log), you'll see that there is not a vertical line at the edge of the frequency band, but a fairly steep slope (24dB/octave is common).

Frequency splitting is a neatly automated tool, but you could do the same thing yourself with some steep highpass and lowpass filters.
 
Last edited:
You must log in or register to reply here.
Back
Top