43 microphones were gathered into small groups and tested using four vocalists (two male, two female). The recorded results are collected online so users may make their own subjective comparisons of a single performance on closely related microphones.
Engineers and producers often make microphone choices based on previous subjective experience, recommendations from colleagues, or simple hype within the industry. When faced with the extensive microphone collections of the studios they work in, it would seem difficult to make careful subjective comparisons of all the available options within the time and budget constraints of an average session. The resources available in this research carefully gather single performances on small groups of related microphones. Hopefully, this affords the user an opportunity to carefully compare timbral characteristics of closely related microphones (and options) without their perception being colored by differences in the talent's performance, and be able to do so on their home computer or mobile device without using up precious studio time. The microphones represented here are a small cross-section of microphones from the UMass Lowell cabinet. The website is designed to assist student engineers in their pre-production session planning, and to encourage them to listen for subtle differences between transducers before and during their sessions.
The initial recording sessions took place over a three-day lockout at the studios of the University of Massachusetts, Lowell. The recording sessions utilized 12 "stations", each containing a group of 3-4 microphones plus a consistent control microphone that would move between each station along with the talent. This allowed each performer to provide 12 solid performances over the course of 60-90 minutes, while collecting recordings at 30-40 different microphones1 over the course of the session. This would also provide comparisons of the exact same performance across closely related microphones within each group. In some groups, comparisons were made between different polar patterns on the same microphone model, or between microphone options (like the DPA Acoustic Pressure Equalizer, and the front vs. back comparison of the Royer R121.)
The microphone array configuration employed in each grouping was developed with respect for understood propagation patterns of the human voice (Olson, 1955), the on-axis polar patterns of each transducer, possibilities of physical placement, angle of sound impingement (Eargle 2004) and consideration of the potential diffraction effects caused by multiple microphones occupying a small area (Woszczyk, 1989). Each microphone in the array occupies a slightly different position, and is thus subject to receiving slightly different acoustical energy than the other transducers in the group. To determine the "maximum" amount of microphones that could be used at each station, arrays were tested in configurations of three, four, five, and six similar microphones (once with DPA 4006s, once with Shure SM-58s) with talent reading from a distance of 12". Four microphones was determined to offer the most microphones and least amount of coloration.
Each performer was allocated an individual 2-3 hour session. The performer was "produced" through takes at each station, guiding the best possible performance and compiling 2-3 takes from which the research team could later chose. As the talent moved between stations, researchers also moved a “control” DPA 4006 omni condenser, pop filter, headphone monitoring station, two video cameras and two absorptive gobos which were placed behind the performer. Each performer was given a precise location for their performance (12” from the center of the array), and the microphones were height-adjusted so the center of the array was directly in line with the talent’s mouth.
Before each performance, microphone levels were calibrated by means of a 400 hz tone generated through a singe point-source powered loudspeaker, at the same height as the talent and a horizontal distance of 12 inches from the center of the microphone groupings. The output of the speaker was calibrated to 85 dB SPL at the center of each array. Each microphone was routed through a standard amount of studio cabling (mic cable to wall panels through conduit to the control room patchbays) into the 212L preamps2 of UMass Lowell's API Vision console. Levels were carefully metered and matched through a SSL A/D convertor and into Nuendo, where each sample was recorded as 24 bit, 88.2khz .WAV files. These files were later normalized to -1.0 dB peaks for consistent evaluation levels.
The test results are made available online as 16 bit, 44.1khz (dithered) WAV files, grouped by station and performance, for users to subjectively compare. In the case of the musical performances, examples are offered as a stereo mix of a complete arrangement (with some compression and spatial effects added in the style of each genre) for a viable contextual comparison of the microphones qualities. They are also available in pure isolation, with no processing whatsoever. The voiceover performances are also presented with no processing.
While every attempt was made to reduce performance and placement variables, this research provides only a basic overview of timbral differences between transducers. In actual practice, elements of proximity effect, placement, sensitivity, and preamp selection could greatly impact a users choice of microphone. The performance distance of 12” was intentionally chosen to minimize proximity effect, and focus attention on other, more subtle timbral differences between transducers.
1 Some microphones were disqualified from the research if they seemed to not be performing to specification, or were subject to some other technical difficulty (bad cable, etc.)
2 The grouping of small diaphragm omni condensors utilized Millennia HV-3D preamps because of an undesirable interaction with the Earthworks OM-1 and the transformer front end of the API preamps.