Help with hard clipping feedback patch in GEN
I have made this patch with a gun recon filter I have found in this forum.
While I'm getting very interesting sounds the gen patch is clipping hard and I'm not sure why.
can someone spot my problem regarding the above?
another question I have - for the moment I'm using rchoose object to generate random values for the amp and freq tables. how can I distribute something that is less random rather in a certain range - meaning higher value in the lower range or higher value in the middle range or higher values in the higher range.
here is the patch:
Irrespective of what your patch does, one of the commonest mistakes that people make with digital audio is to have the gain set way too high at source.
In the good old days of "noisy" analogue signal processing, the target was to get the signal as close to 0dB as early as possible and to mix and process around the same level to achieve the maximum signal to noise ratios. Most (but by no means all) mixers and signal processors had "headroom" of +6dB (or more) before the signal level "hit the rails" (i.e. reached the supply voltage of the amplifiers) and started serious hard clipping. This was especially true of valves (which is why they were so good for saturation and overdrive effects).
However, in the digital world, where data bandwidth is several magnitudes greater (24bit, 32bit or even 64bit instead of the nominal 16 bit ADAC equivalent on CD's) and where underlying signal noise should be absent, you should actually have the signals running at something more like -18dB RMS or-6dB peak in comparison to an analogue signal. This is usually indicated on digital meters as transitioning from green to orange rather than from orange to red for 0dB analogue. If you overdrive your digital "pre-amps" at source, the transient clipping will start at the nominal 0dB even if you can't hear it and/or your meters/faders seemingly go up to +6dB or more! And further down the line that cumulative clipping will become horrendous. So just try turning the gain down first. You might well find that your patch is not the problem!
The exception is where you have an analogue source being converted to a digital processing system through an ADAC. Here, the source gain in the analogue section should be aimed towards 0dB but the converted digital signal should then be dialled down to -18dB/-6dB.
And NEVER try to "normalise" the signal (expanding back to 0db peak) until you get to the very final post-production mix.
I hasten to add that I realise that you might already know all this but everybody needs to know the difference.
thanks for your informative answer. learned a lot!
A good example of this is my setup. I use an XR18 or X32 rack mixer that can accept analogue or digital signals into any channel. All the outputs can be routed to either analogue or digital (or sometimes both) outputs. Because all the faders and signal level meters are geared up for controlling both analog and digital signals, they have that deceiving trait of having +10dB of headroom built in for the analogue stuff. However, if I feed the digital output of Cubase into channels next to an analogue source and try to balance them both around 0dB all the way through the chain the beautifully clean Cubase signal starts clipping. Answer - turn down the final output of CuBase and use the amp in the mixer to bring it back up again at the digital-analogue interface.