Microphone recommendations

Ease of use. Built-in microphones beat USB microphones beat XLR microphones with an interface, in that order. More cables and boxes mean a longer setup, more to carry, and more to forget. Compare your phone camera to a dedicated camera with a bag of lenses: which one do you actually use more? A modest microphone that you have always with you, and that requires less time to setup, can be more useful than an excellent one in a drawer.

Flat frequency response. A spectrogram tells the truth about what the microphone passes through. Vocal-recording microphones deliberately shape their response (a gentle bass roll-off, a presence peak around 5 to 10 kHz) to flatter a voice for radio or studio recording. That works for recording. For analysis it adds a coloration to every reading. Measurement microphones aim for flat instead, and that is what an analyzer wants.

Extended high-frequency response. Ordinary microphones roll off near 20 kHz, which fits the 44.1 and 48 kHz sample rates used in everyday vocal work. For partial analysis of bowls, gongs, and instrument bodies, where useful content sits above 20 kHz, you want a microphone whose response reaches 30 or 50 kHz. Anything beyond that needs laboratory-grade instrumentation or ultrasonic transducers, and is outside what this page covers.

Omnidirectional vs cardioid. Measurement microphones are omnidirectional: they pick up equally from all directions, which keeps the frequency response clean and predictable. Cardioid microphones reject sound from behind, which is good for rejecting room noise but adds proximity effect and off-axis coloration. For analysis work that cares about the actual sound, omnidirectional in a reasonably quiet room is the better default.

Individual calibration. The miniDSP and Earthworks microphones in the recommendations below ship with a calibration file specific to that unit. If you care about absolute levels or formant amplitudes the calibration file is the difference between a good measurement and a precise one. For relative work, looking at how a single voice changes over time, it matters less.

The laptop's built-in microphone needs no external device and no cable at all. Select it as the input and record.

The Samson Go Mic and both miniDSP UMIK models are USB devices and do not need an extra audio interface. A single cable runs from the microphone directly to the computer. Plug, select, record.

Earthworks measurement microphones, and any other XLR microphone, need a separate USB audio interface that supplies phantom power. An interface also helps when you want to run at sample rates above 48 kHz: built-in laptop inputs are usually capped, and a dedicated interface gives you a quieter preamp and stable timing into VoceVista. This requires two cables and an interface box instead of one cable: XLR from the microphone to the interface, USB from the interface to the computer. What to look for:

A good microphone in a bad room gives you a clean recording of a bad room. Before upgrading the microphone, look at where you are recording.

Distance. For vocal work, a hand's width to an arm's length from the mouth is a reasonable default. Closer gives you more direct voice and less room; further gives you more room and a fuller, more natural tone. Watch the spectrogram while you move and pick the distance whose result you like.

Angle. Omnidirectional microphones do not care much about angle. Cardioid microphones do: aim at the source, and expect the sound to dim and dull as you go off-axis. For measurement work with an omni, point it at the source anyway so you can repeat the setup later.

Room. Soft furnishings, bookshelves, curtains, and rugs absorb reflections and quiet a room down. Bare walls, glass, and tile add reflections that show up as a thicket of vertical streaks on the spectrogram. A quiet, soft room with a modest microphone often reads more clearly than a great microphone in a reverberant space.

Monitoring. Use headphones for live monitoring during analysis. Loudspeakers feed back into the microphone and confuse the spectrogram with delayed copies of what you just sang.

Input level. Set the input level so the loudest passages sit comfortably below clipping, but well above the noise floor. Too quiet and the spectrogram shows the preamp's hiss more than the source; too loud and clipping adds spurious high-frequency harmonics that were never in the original sound. VoceVista shows an input level meter while you record. See the quickstart guide for the procedure.

Consistency. Most VoceVista work compares two recordings: yesterday versus today, before versus after, this take versus that take. The comparison is only as good as how alike the two setups are. Same microphone, same distance and angle, same room, same gain. A modest mic kept in exactly the same place every session tells you more about real change than an excellent mic moved a few centimeters each time. For instrument and bowl measurements you might want to repeat weeks later, write down the distance and angle, or take a photo of the setup, so you can rebuild it. Repeatability beats absolute accuracy.