[music-dsp] wow/flutter tape delay
James Chandler Jr.
jchandjr at bellsouth.net
Mon Sep 9 12:44:58 EDT 2002
> I am putting together a simulation of a tape delay and am currently adding
> wow and flutter (speed/pitch variation) but was wondering if anyone knew
> the typical modulation shape and frequency range (would this modulate
> aswell?) of this effect?
For wow, I'm guessing that the typical mod frequency might be related to the
circumference of the tape capstan (the cylindrical piece that drives the
tape) versus the tape speed. Capstans were pretty huge on big expensive tape
machines, but the expensive machines didn't have much wow and flutter
Just a wildly inaccurate guess (memory fails), the old echoplex either ran
at a nominal 3.75 or 7.5 inches per second, and the capstan was about a half
I think flutter was a combination of several factors, among them-- Tape
"stiction" on the heads and other components, where the tape repeatedly
"grabs" then "releases" at a fast rate... Random gunk on the capstan or
pressure roller... Poor regulation of tape pressure against the heads,
causing the amplitude to rapidly (and only slightly) fade in and out...
A feature I use in my Flange and Echo/Chorus plugins to "kind of"
approximate wow/flutter, is a "Random" parameter in the LFO, calibrated from
0 to 100 percent. Rather than constantly jiggle the LFO wave over each
entire LFO cycle, if the "Random" control is non-zero, I recalculate the LFO
rate with the Random factor each time the LFO finishes a cycle and crosses
zero. So each individual LFO wave period is "steady", but none of the LFO
wave periods have exactly the same duration.
If the "Random" control is at 100 percent, ferinstance, each LFO wave period
will be randomly selected somewhere between the Base LFO rate, and 2X the
Base Rate. At 50 percent, the wave periods will range from the Base LFO Rate
to 1.5X the Base Rate.
In my implementation, the LFO amplitude is lower, if it is randomly set
higher than the Base Rate, so the depth of pitch modulation randomly varies
along with the rate of pitch modulation.
There are lots of fancier ways to randomize LFO's, but this is pretty
simple, and makes a pretty "organic" wobbly result.
Additional useful controls would be Lo Freq and Hi Freq Rolloff. I've found
that a simple first-order tunable HiPass for Lo Freq Rolloff, and a simple
first-order tunable LoPass for Hi Freq Rolloff, gives a pretty musical range
of control, considering that they are such simple controls. You would insert
the Rolloff filters directly ahead of the mixed delay line input, so that
both new input and feedback input will pass thru the Rolloff filters. With
high echo regeneration, each successive echo gets "thinner" as it tails out.
You wouldn't apply the Rolloff filters to the output mixed dry signal, of
Years ago talked to the old engineer who invented the Echoplex. He said the
upper mids are deliberately boosted, and the lows and highs deliberately
attenuated, to "make it sound good on guitar." While designing the device,
he had famous Nashville studio guitarists beta-test it, and kept fiddling
with the frequency contour until they were happy.
They did a good job, IMO. A "thin" echo seems to be generally more musically
useful than a full-range echo. Full-frequency-range echos are often too
dupswapdrop -- the music-dsp mailing list and website: subscription info,
FAQ, source code archive, list archive, book reviews, dsp links
More information about the music-dsp