|
The Frequency Isolation Thread
|
View this Thread in Original format
| StephenWiley |
The basics.....
Sidechaining, filters, roll offs, panning
Saw another thread about Frequency Sweeping....Looks very interesting.
Anybody have some important pointers here on frequency isolation ? |
|
|
| Knowland |
All filters are imperfect. It's a balance of slope, phase, ripple, and ringing. Thing to remember is, so is almost every physical sound that occurs. Digital sound is physically occuring through your speakers with digital artifacts. Filters actually make things sound less digital sometimes.
Amplitude effects are a linear process, to a point. If you drop the amplitude of a wave at 60hz, over 6 samples, you get a discontinuity because the decline was faster that the periodicity of the wave. That causes a bump in the sound. To get a smooth sound, volume envelopes, including compression, should not rise or decay faster than 2 periods of the lowest frequency present in the signal.
Panning has a lot to talk about. It's best done on loudspeakers I think because you hear one side in both ears, HRTF crosstalk. Even if you pan a sound at about 500hz to the hard side, if a similar frequency is on the other side, you lose the positioning of the sound because the phase information will be masked by the second signal. To counteract this you must decrease the volume of the other side to overcome the masking, or else it is stereo wash! So when it comes to positioning n the stereo field, keep one frequency dominant on the right side.
I'm not like a pro sidechainer or anything. For frequency isolation, it can be used to duck for a sound that would be masked normally. If that fits with what's happening in the audio, use it. Sidechaining is only more useful than a normal compressor because you don't have to pass through all the material, you can mute the sidechain. |
|
|
| kitphillips |
Generally, the high pass filter is your friend. It can clean up a mix like nothing else. Generally on live sources, you may want a low pass filter too, you can almost ALWAYS low pass from 18k or 10k if you need to create space. And you should never underestimate the power of a boost in the presence at above 8 k, theres often a lot of space up there that doesn't get used enough, especially in vocals and pads...
When your carving out the space you want for each sound, you probably want to shave off the highs and lows more than anything else. I've really found lately that meddling with EQ in the midrange of a sound isn't that helpful. Usually you want to clean up the extremes of the frequency spectrum, not play around with the body of the sound too much;) |
|
|
| Subtle |
| Low cut everything that isnt suppose to be bass in all tracks even if its not a bassy sound. |
|
|
| Eldritch |
| quote: | Originally posted by Knowland
Digital sound is physically occuring through your speakers with digital artifacts. |
I'm not sure what your point is, but there's no inherent artifacts in properly sampled digital audio. |
|
|
| Knowland |
| quote: | Originally posted by kitphillips
Generally, the high pass filter is your friend. It can clean up a mix like nothing else. Generally on live sources, you may want a low pass filter too, you can almost ALWAYS low pass from 18k or 10k if you need to create space. And you should never underestimate the power of a boost in the presence at above 8 k, theres often a lot of space up there that doesn't get used enough, especially in vocals and pads...
When your carving out the space you want for each sound, you probably want to shave off the highs and lows more than anything else. I've really found lately that meddling with EQ in the midrange of a sound isn't that helpful. Usually you want to clean up the extremes of the frequency spectrum, not play around with the body of the sound too much;) |
Those are good tips there. When I EQ the body of a sound I like to use low Q bell type filters, or if you use FL Studio those nifty 4 order bells that make the bell curve flatter. Q of .1 is good as it adjusts the entire octave. |
|
|
| Knowland |
| quote: | Originally posted by Eldritch
I'm not sure what your point is, but there's no inherent artifacts in properly sampled digital audio. |
That's true but when you start editing samples its easy for beginners to overlook the incongruencies caused by chopping and breaking the continuous signal that was originally sampled. Try generating a sine wave in a program, that does not apply a ramp up envelope. This causes a bump in the frequency it went from 0 to sin() = 1 in one sample, with no "acceleration" into it. This is why some people including myself right now say having a band pass or low pass filter on a sine wave makes it sound better. |
|
|
| Knowland |
Just because I'm bored I make a picture to illustrate this, and a sound clip so you can hear it.
digital_wave_bumps_example.wav - Hosted on SaveFile.com
It's a 55hz tone processed with 24db filters. The first tone is band passed with a basspand of .166/octave, the second is lowpassed, the third is untouched digitally generated, the fourth is enveloped on the front and back over two wave periods.
To be honest the FFT process exagerates things because it is a filter with trade offs. It used a 8096 size fft. Those bumps might not really exist in the fourth example. The bandpassed tone produces a bit of ringing below the 55hz, but on the fourth tone it gets even better frequency performance and no ringing.
This is really only important in synthesis, as when you generate your number set with a sin(), it makes a true sine wave but it does not create the proper starting and stopping wave. Hence no physical simulation with such simple generation. |
|
|
| wrzonance |
| quote: | Originally posted by Subtle
Low cut everything that isnt suppose to be bass in all tracks even if its not a bassy sound. |
Seconded! This was trick number #1 for reclaiming the low frequencies from your drums in recording school.
Hats, Rides, Snares, even Toms have no useful low frequency sounds.
But microphones record EVERYTHING (they're not like your ears/brain which automatically filters sounds for you)... so it was very common to either apply 12 db of a low shelf (shelving filters cause less octave distortion), or a high pass filter with a soft db/oct curve.
:)
As far as SPL in the frequency spectrum, low frequencies dominate. So you need every last drop of low frequencies available for your instruments that ACTUALLY MAKE USE of that part of the spectrum.
Word up.
---Adam
NOTE: *Most* professional sample libraries will have already applied this sort of processing to their sounds. However any SYNTHESIZERS or RECORDED sounds you use, won't (duh). |
|
|
|
|
