Results

Preference Ranking

The preference rankings need some discussion. The listeners had to use their own intuition and “make up” appropriate numbers using internal, non-verbal criteria. They repeatedly compared the “original” sounds with the recordings, noting that there was significant loss between the sounds of the singers and the sound of any microphone through the Genelec, which is quite flat (+/- 1.5 dB on-axis from 100 Hz. to 20 KHz.). This is an important point: the difference between the actual voice and the “best” mic’s rendition of it is significantly greater than the difference between the “best” mic and the “worst” mic, confirmed by both listeners’ comments and the data. As the listeners assigned scores, they were listening to sets of four microphones, so that their rankings and preferences in any given test were influenced by which other microphones were in the test set.

The next qualification to keep in mind is that the rankings were very dependent on the individual voice and its context. Each mic recorded each of the five singers both a cappella and with a pop music mix. On several occasions listeners queried whether a mic had its letter ID changed (it never was) when they began listening to a new vocalist.

We must also consider the relationship between subjective preference and frequency response. Voice recordings are strongly dependent on the microphone response characteristics between 1500 Hz. and 10,000 Hz., and microphones with boosts or peaks in this region tend to be preferred over flat microphones, probably because of the increased intelligibility and prominence of consonants, sibilance, and our hearing sensitivity in that region of the spectrum. The B&K 4011 was demonstrably the “flattest” microphone in the comparison. It elicited numerous comments such as “muffled” and “dull” and was low in the preference ranking. This does not mean that “flat” is bad – it simply means that when recording pop vocals, frequencies between 1500 Hz. and 10,000 Hz. should probably be judiciously boosted. Any of these mics can be EQ’d to your heart’s content, while we didn’t use EQ . This is a matter of art, not science; intuition, not formula.

With that said, here’s a chart ‘n data showing listeners’ preferences:

Microphone	Mean Score	Standard Deviation
AKG C3000	6.2	.62
Audio-Technica 4033	7.1	.5
Audix OM-3xb	6.3	.97
Beyerdynamic 834	6.5	1.3
Bruel & Kjaer 4011	5.5	1.1
Neumann TLM 193	4.7	1.2
Sennheiser MKE 4032	6.2	.78
Shure SM58	5.3	.86

Now these are pretty interesting results, and not what I expected. The standard deviation range indicates the amount of consensus among the listeners – a wider range, like the Beyer has, means there was less agreement about this microphone, while a small range, such as the Audio-Technica’s, suggests substantial agreement among listeners. So, we can predict preference for the Beyer to be variable while preference for the Audio-Technica will be comparatively consistent. Another caution: differences in preference scores of less than a point are insignificant, so that in reality there are two general sets of preferences: the more preferred microphones (AKG, Audio-Technica, Audix, Beyer and Sennheiser) and the less preferred microphones (B&K, Neumann and Shure).

The next thing to do is have a hard look at the response curves to see what, if any, correlations there are between response and preference. The primary insight has to be that flat response is not desirable for recording pop vocals. More specifically, some amount of comparatively broad and gentle boost in octaves 8 and 9 (3 to 10 kHz) seems to be preferred, and flat low-frequency response is also slightly preferred. There is a sort of family resemblance between the Shure, the Sennheiser and the Audix curves, with low frequencies rolling off and peaky boosts from 2 to 10 kHz. The Sennheiser and Audix obtained equal preference scores, while the Shure was a point below. I speculate that the Shure’s narrower peaks and poorer extension into the tenth octave accounts for listeners’ comparative lack of preference for it.

The failure to prefer the Neumann and B&K is a stunner for me – it challenges the bedrock notion of “accuracy” and the desirability of flat frequency response. These are really excellent microphones, highly suitable and desirable for a broad range of recording applications. However, the listeners’ preferences suggest that recording pop vocals is not a particularly good application for these mics. The Beyer and the Audio-Technica also have quite flat response, but I speculate their slight boosts in the high frequency range lead to their preference rating. The Audio-Technica’s slight boost below 10 kHz and significant peak above 10 kHz may give it a particularly desirable “airiness.”

COMMENTS

Post a Comment