One thing that would be fun to be able to include with Kingfisher is a fiddle delay. So I tried making one! It seems like it should be pretty simple: take audio in, wait some number of milliseconds, then send it back out. With some playing around at home, it seemed like a delay that matched the current tempo and repeated things three times could be good. I coded something up, and it seemed to work ok.

I finished it just before we went down to play CDNY, and I brought it along. Unfortunately, this is a dance with a relatively short amount of time for setup: the hall opens at 6:45 (though we were able to get in a bit before that), the beginner's workshop starts at 7:00 (though the beginners were, as usual, a bit late, and we had extra time), and the dance starts at 7:30. This is very tight; if I'm not running sound I try to have an hour for setup (6pm arrival before a 7pm workshop) and it usually takes ~40min to get everything set up and sound checked. Which all meant that the delay was not a priority, and while we did line check it visually during the workshop (when you can't be noisy) we didn't get farther than that.

Unfortunately, it didn't work out. We tried checking it during a walkthrough, and then turned it off as soon as the set started. Which was fine, and the set was good—I suspect none of the the dancers even noticed! But here's what it sounded like, with the fiddle mic panned hard left and the delay hard right, though it wasn't this loud. I turn the delay on at 0:07:


There are several things going wrong in that clip, and I'm not sure how much each is responsible. The big one is that the fiddle mic (AT Pro-35 clip on) doesn't only pick up fiddle: it also picks up the monitor and general hall noise. Normally this is a nuisance, because it can lead to feedback, and it makes the overall sound mildly muddier than it would normally be, but it's worth it because a mic gives you the best fiddle tone to work with. With a delay, however, it's now much more of a problem: no one wants to hear each of the caller's words four times.

Another issue is that this particular hall is a rectangular box, full of hard reflective parallel surfaces. This means you have a lot of resonance. You don't want zero resonance (I've played in a hall like that, also in NYC, and all the energy from the dancers is just swallowed up by the perfectly treated walls and ceiling) but this was more than ideal This probably meant there was much more caller, monitor, and other sources of non-fiddle sound going into the fiddle mic.

Listening, there's also some sort of accumulation of high frequencies. At the dance I thought the problem was that I'd set the input level wrong on the fiddle delay and we were getting some sort of digital clipping, but that's not it. Then I thought maybe the delay is picking up its own output at a very low level, but I don't hear this playing at home. It also goes away when the hall quiets down.

It would be useful to know how much of this is that this is just not going to work with an acoustic fiddle and mic (and you just need a pickup or solid-body if you're going to go this direction), vs being very unlucky with the hall we tried it in (most halls will be much less resonant), vs my coded delay not being good enough (and maybe using a commercial one with tempo tapping would be better). My next step is to play around with this more at home, trying it a clip-on fiddle mic and higher volumes, but I'm curious whether anyone has a clearer idea of the problem from listening to the clip?

New Comment