Two nights ago I had to record a Japanese taiko drum ensemble. I was using two DPA 4011 mics, but I got a fair bit of distortion even though my recording levels never reached 0dBFS; my actual peak was ‑3dBFS. Why was I getting this distortion/compression, and how can I mitigate it in the future? It looks like I’ll be doing this again soon!
SOS Forum post
SOS Technical Editor Hugh Robjohns replies: You say you were recording an ensemble with two mics, which implies a stereo array and therefore not close miking, which obviously reduces the SPL at the mics significantly, but even so the transients would be big. So transient overload in the mic impedance converter is a possibility, and for that kind of gig selecting the pad would be a wise precaution. A high‑quality capacitor mic will capture brief transients which dynamic mics would round off, and which some meters won’t show.
The next and probably most likely possibility is overloading the front end of the mic preamp. Sensitivity of the DPA 4011 is 10mV/Pa, so if the transient peaks reached 120dB SPL at the mics, which is plausible, the output signal would be around ‑12dBu, which is approaching line level and might be getting a bit hot for some preamps, especially if there’s a chunk of gain dialled in. Depending on the circuit design it’s often possible to overload the front end of a mic preamp even with the gain at minimum. Input pads or inline attenuators are the saviour here, although the mic’s internal pad may be sufficient to keep the mic signal within practical bounds.
Finally, sample‑peak metering and intersample peaks within the DAW could be the culprit. Sample‑peak meters commonly under‑read on transient‑rich material, often significantly, particularly if working at base sample rates. Under‑reads of 6dB or more are not unusual. When replaying your recording it might be worth pulling the master fader down so the DAW output meters indicate peaks of ‑10dBFS just to see if the distortion is due to intersample peaks hitting the D‑A converter in the replay chain.
When recording anything, I get very nervous if peaks reach ‑6dBFS... Peaking to ‑3dBFS is just asking for trouble...
But if the distortion remains, it was a problem with the recording chain overloading somewhere — there is a declipper module in iZotope RX which works very well, and will probably salvage your recording to an acceptable degree. Personally, when recording anything (and relying on sample‑peak meters) I get very nervous if peaks reach ‑6dBFS and I aim to keep them around ‑12dBFS or lower. Peaking to ‑3dBFS is just asking for trouble... With strong transient‑rich material I’d be allowing even more headroom with peaks to around ‑15dBFS or so. The average signal level will be much, much lower than the transient peaks.