another clipping issue (or is it?)

[standage]

on a recent session, on the drum overhead tracks (AKG 451s into TL Audio pre --> TL audio compressor into a Digi 003 rack) i got some nasty clicking.

not 'that' nasty, but some of my stick hits (i'm playing and engineering myself, which i loathe) sounded really pronounced so I zoomed into the max, and see that the wave form goes from positive to negative in 0ms. So obviously that's going to make a bad noise. But the signal wasn't too hot on the pre amp, it got compressed anyway, and the input level to the interface wasn't hot either.

So, I re-compressed the track (using the same outboard), thinking it would reduce the 'spike', but it made it bigger. In every other case, the waveform's loud bits were quieter and the quiet bits louder (hurrah for compression) except these anomalies, when the 'positive to negative in 0ms' came out more pronounced. And to make things worse, a software compressor does the same thing (to the sound that is. I haven't checked out the bounced waveform).

Please offer your suggestions, because it's driving me nuts OR check out my crack pot theory: the fastest attack time of my compressors (outboard and plug in) is 1ms. Is this the signal freaking out the diaphragm BEFORE the compressor has time to react?

 

[SouthSIDE Glen]

Please offer your suggestions, because it's driving me nuts OR check out my crack pot theory: the fastest attack time of my compressors (outboard and plug in) is 1ms. Is this the signal freaking out the diaphragm BEFORE the compressor has time to react?

There's nothing crackpot about it. Remember that when digitally recording at 44.1kHz, a full 44 samples go by in a millisecond. On can fit a lot of digital transients into a 44-sample range if they really wanted to (not that most of us would ever want to .)

If re-tracking is not an option, I'd get rid of those transients with manual waveform editing - Which is always how I attack the worst transients anyway; compression should be (IMHO) used to control the signal, not the noise.

As far as what's causing the clicks to begin with, there's insufficient information at this point. When you say "overheads" in the plural like that, do you mean that you are recording stereo OHs and are getting the clicks in both tracks? And are they being recorded simultaneously with your other tracks (e.g. snare, kick) through the same Digi interface? Have you tried rotating mic assignments? And so forth.

G.

 

[standage]

thanks for reassuring me. now that you've confirmed my theory, it almost seems OK to go in and edit the waveforms manually. It's not that I mind the work (i sometimes find it quite therapeutic, like before a bought a drum triggering plug in and inserted my samples manually, one by one), it's just that I figured there would be a better way to fix it.

to get to the bottom of the cause: yes i had stereo overheads, a left and a right both giving the same clicks, but i guess I might have been running them hotter than i thought, and just compensating with the compressor. now i guess i should level them with the compressor off and then turn it on to track.

thanks for your reply.

 

[mixsit]

..So, I re-compressed the track (using the same outboard), thinking it would reduce the 'spike', but it made it bigger. In every other case, the waveform's loud bits were quieter and the quiet bits louder (hurrah for compression) except these anomalies, when the 'positive to negative in 0ms' came out more pronounced.

Sharper than other drum hits, random and not related to legit loud spots?
I'm thinking this looks a lot like clock or software glitches.
Especially if they don't correspond to transients on the snare or other tracks.

 

[standage]

no, each spike relates to a cymbal hit. the snares came out fine. which is odd, since the snare is way louder, and the overheads weren't particularly close. i should post an audio example....

 

[Ripthorn]

I agree with glen, I usually do this sort of stuff by manual sample editing. It's really not bad if your DAW allows it and if you remove one or two samples of nastiness, you get rid of the clicks and most people will never even know they are gone.

 

[SouthSIDE Glen]

i guess I might have been running them hotter than i thought, and just compensating with the compressor.

Yeah, if you were running too hot into the compressor and you're getting transients faster than the compressor's attack clamping capability, the transients will sneak by and come out too hot.

Also to watch when recording with compression is the output gain setting. Even if you're sending good levels into the compressor, if you have the output (make-up) gain set too high, that can drive those transients into clipping as well.

Now that you mention that you're getting the clipping on each cymbal hit, I'd add a caveat that common sense of course should come into play as to how to handle the transients in post. I am still a fan and believer in attacking rogue peaks with editing rather than a dumb circuit or algorithm, but of course if you have a couple of hundred "rogue" peaks over the course of the song, manually editing each one individually may not be a reasonable expectation time- and effort-wise. But if we're only talking a handful to a couple of dozen or so (it's a judgment call), manual editing would remain my recommendation.

What if there are a couple of hundred or some other quite large number? Personally I'd call that a failed recording and go back and re-track. Only if that were not possible (the drummer moved to Zimbabwe and took the drum set with him, or some other *legitimate* reason for not re-tracking that doesn't amount to laziness) would I throw a look-ahead compression plug that can handle the transients at it.

IMHO YMMV SNAFU ETC.

G.