Geoff Grace wrote on Thu, 22 April 2004 13:51 |
In my opinion, you deserved a much better send-off. |
zakco wrote on Thu, 22 April 2004 16:50 |
George, it's great to see you hear. -Z- |
Nika Aldrich wrote on Fri, 23 April 2004 12:48 |
...Whew! OK, I made it in. Anyone else still stuck out there? Anyone seen NYC Drew or Alphajerk or Curve yet? Nika. |
Stephen Fortner wrote on Fri, 23 April 2004 22:00 |
Any opinions expressed here are entirely my own. They are not intended to reflect the opinions, editorial policy, or official position of any publication, media entity, manufacturer, or other organization with which I may be associated. |
Nika Aldrich wrote on Fri, 23 April 2004 14:33 | ||
Such as? |
chrisj wrote on Sun, 25 April 2004 01:42 |
Pleased to meet you- I never got around to spending much time in the EQ forums, and when you suggested I call you and chat, I talked myself out of it Maybe this will be a better way. Power of the internet and all. I'm keeping busy in the same way that attracted your attention, as always- for today, it was rehacking a slew clipping routine so it became a soft clip- if threshold is X and going beyond the threshold is X+Y, it became X + sqrt(Y). Yesterday it was setting that threshold to within X of the previous sample. Formerly (embarrassingly, on CAPE and the last WOMP) it was having the threshold clipped to zero instead of to the previous sample- ouch, crunchy... Got a hell of an improvement off even the hard clipping version, surprisingly one of the primary things it does is define bass better. The soft clip version REALLY defines bass better. And the effect in the highs is as intended very much like acoustic distance being present. I wish I'd had this debugged even a month ago... wonder if I should be coming up with a test file, seeing as the forum appears to be allowing the uploading of a file? Anyway, working- and more than happy to belatedly chat, better late than never. |
George Massenburg wrote on Tue, 27 April 2004 18:29 |
Well, the SQRT function is the right way to go, but I wonder if it should be more continuous - there's clearly a more-or-less sharp break even for nominal values of Y. I'd suggest a continuous function - perhaps cobble one together from a Taylor series expression. You didn't mention how you were fabricating this? Is it in data domain? Or analog? Thanks for the stimulating thought, George |