I started with a WAV file of full scale Gaussian white noise. Spectral analysis reveals that it’s very flat, as it should be. Then, I piped it into the Korg MR-2’s mic input, set to Low sensitivity and 0 dB level. The graphs labelled “direct to Korg” show that it has flat frequency response from 20 Hz to 20 kHz, which is great. The graphs labelled “dead kitten” refer to the Rode Stereo VideoMic Pro with the dead kitten windscreen installed. Actually, the frequency response is similar with or without the dead kitten. Maybe the foam cover alone gives it the muffled characteristics. The Rode shows bass roll off below 60 Hz, which is expected, but also that frequencies above 8 kHz fall off by at least 6 dB. The dips every 1000 Hz in the first test were a bit puzzling. I thought it may indicate the dimensions of my room, which could behave as a resonant chamber. But 1 kHz implies features on the order of 1 ft in size. The only thing that’s around 1 foot is the distance between the two drivers on my Sonus Faber Grand Piano speakers. That’s still not a satisfactory explanation, since they were equidistant from the mic. To try to remove room acoustics or background noise, I placed the mic right up in front of the speakers and ran the last test. That appeared to boost the low/mid response.
These results show that the Korg MR-2 has great frequency response and that the Rode Stereo VideoMic Pro definitely needs equalization in post processing to get rid of the muffled sound.
Next, I evaluated the Canon 5D-MkIII’s ability to record white noise. The high frequency response is fine, flat up to 20 kHz, but the low frequency begins to fall off as high as 150 Hz. That’s no good for recording music. I’m sure the high pass filter is built into the camera to help avoid clipping. This is another reason to not use the camera’s onboard recording.