There are two reasons why one might split the frequencies of a signal.skyhigh wrote:when would it be a good idea to split frequencies?
1> To maintain control of the signal.
2> For creative purposes.
-------------
If we split the signal for control purposes, we are most likely looking at multiband compression. Low frequencies trigger compressors before higher frequencies (although I actually don't know why
)...By splitting compression into multiple frequency ranges (bands), we can process bass with more precision than single band compression. This will give a clearer mix, since using frequency dependant compression can allow for the 'loudest' frequncies to be attenuated (turned down) whilst other frequency ranges are unaltered and thus true to what was orignally intended.
-------------
Frequency splitting for creative purposes follows the same principal as using frequency splitting for control purposes. It gives greater control over what happens to both the signal and the overall mix.
Applying effects to separate frequency ranges allows for more precise control. As an example, we might compress the lowest frequencies, whilst adding chorussing to the midrange, and saturation to the highs. Doing so would give a signal that was present, yet mixable. The lows wouldn't overpower any higher frequencies in the mix, the midrange would sound 'interesting' whilst the highs would sound 'present'.
Sometimes it is interesting to add the same effect across all the frequency bands, yet with different settings for each band. Apply distortion would be a fine example of this scenario.
-------------
Splitting is done in two ways.
1> By using filters (phase cancellation)
2> By subtracting a copy of the signal from itself.
In both cases, we would send a copy of the signal to a bus, upon which, a filter or maths unit would function to split the incoming signal into bands.
Using filters is the easiest way.
To split the signal into three bands, we would send the signal to three buses. On each bus, we would use a filter to separate the signal into bands.
-On the first bus, we would use a low shelf filter to remove all frequencies above the cutoff point. This becomes our low frequency band.
-On the second bus, we would use a band pass filter to only allow the mirange frequencies through. This becomes our midband.
-On the final bus, we would use a high shelf filter to remove all frequencies below the cutoff point. This becomes our high frequency band.
Following the filter on each band, we could add compression or effects.
-------------
Of course, I could be chatting shite, so feel free to add your experiences of frequency splitting too!
Pat
Perfect-reconstruction filters are all well and good, but as soon as you start processing (the comp starts working) you have phase problems at the crossovers.
