posted 12 January 2002 23:43            

chorus_flange.kym

The first is a chorus/flanger made up of two consecutive delay lines with independent delay length controls. The signal first passes through a high pass filter with adjustable cutoff. Next it flows into the first delay line that is modulated by a sinewave LFO. From there it goes into the second delay line that is modulated by the same LFO to varying degrees from none to full on like the first delay line. Then some of the signal is tapped off and fed back into the input high pass filter. The sign of the feedback can be either positive or negative. Different comb filtering occurs in those two cases. Finally the output of the delay lines is mixed back with the original signal in a Wet/Dry ratio for final output.

I was impressed at how deep the flanging effect is, and I liked the sound of the high pass filter.

There is a stereo version of this chorus/flanger provided here that is so incredibly deep you have to really back off on the wet/dry mix to bring things back down to the ground... The Stereo version requires only 2 DSP's to run.

Finally, there is a triple-bandpass delay with feedback, but with much longer delay values than for chorus/flanging. This one is intended to provide frequency selective echo delays for up to 1 second long. I have the delay sliders in the VCS set up so that integer values correspond to 1/16th notes at BPM. This triple filter/delay is built from a class object included in the Sound file. That class object merely encapsulates the bandpass filter and the delay line and its feedback. The class prototype is included for your viewing. Putting this stuff into a class object actually cuts down on the number of DSP's required to run it. This stereo 3-band/delay requires only 2.5 DSP's to run.

The 3 channels in the filtered delay are numbered 1, 2, and 3. Channel 1 is tied to the left channel input but can be panned anywhere on output. Channel 2 is tied to a mono mix of the input and can also be panned anywhere. Channel 3 is tied to the right input channel, and of course, can be panned anywhere. All three channels have independent bandpass filtering of their input. The feedback on each channel feeds from the output of the delay line back to the input of the bandpass filter.

Enjoy!

- DM

posted 31 December 2002 15:07            

Anyway what is the theory behind the split delay line? what do you gain?
I know that the old bucket brigade flangers had an advantage over DSP in that the delay was changed by changing the sample rate of the whole circuit instead of skipping samples (as the DSP
does).
Even linear interpolation doesn't help if the delay change requires a jump of greater than one sample. Does your new technique help mask this problem by making each delay stage jump by
less than one sample therefore allowing the delay as a whole to jump more than one? If this were the case, then maybe more than two delays would be better, and the modulating signals could be the same.

Am I on the right lines here or have I missed the point?

posted 11 January 2003 22:10            

chorus_flange.kym

I just checked in and found your queries. Here is the missing Sound file. It had been removed because at one time there was a rather severe limit to the number of submitted Sounds that could be made. That limit has now been removed.

As for why there are two cascaded delays in the design....

1. This was the specification from those German DSP pro's and I merely copied the specification

2. The total delay can be written as

Dtotal = (D1 + M(t) + D2 + f*M(t)) * 10 ms.

where 0 < D1 and D2 < 1, modulation -1 < M(t) < 1, and fraction 0 < f < 1. This is equivalent to a single delay controlled as

Dtotal = (D' + M'(t)) * 20 ms.

where D' = 0.5 * (D1 + D2), and M'(t) = 0.5 * M(t) * (1+f)

So why do this? I can only speculate that perhaps it is a way of reducing the effects of quantization error since the computation is performed in two stages with coefficients that are roughly halved in their effectiveness?

Good question.

Cheers

posted 18 January 2003 03:14            

I have something for your eyes only (at first).
I used your old Email address dmcclain@azstarnet.com but it bounces
everytime. Can you send your new one at c.agelauff@freeler.nl