Recording with an audio interface: what it does and how to set it up

A microphone produces a very small electrical signal. Something has to amplify it and then turn it into numbers a computer can store. Those two jobs are the preamp and the analogue-to-digital converter, and an audio interface is a box containing both, connected over USB.

Your computer already has both, which is why the built-in microphone works. They are compromised by being inside a machine full of electrical noise and built to a price that assumes nobody is recording anything serious.

An interface also gives you XLR inputs, which is what most good microphones use, phantom power for condensers, physical gain knobs, and headphone monitoring with no delay.

What it does not do is make a bad microphone sound good, or a bad room sound treated. It removes a ceiling rather than raising a floor.

Most interfaces are class compliant, meaning they work without drivers on macOS and modern Windows. Plug it in and it appears in your sound settings. Some, particularly on Windows, want their own ASIO driver for lower latency, and that is worth installing if the maker provides one.

Set it as both input and output in your system settings, then again inside your recording software, which often keeps its own separate device selection.

Plug headphones into the interface rather than the computer. This is the point of having one: you hear the microphone directly through the hardware with no processing delay.

If a condenser microphone produces nothing at all, you have almost certainly not switched on phantom power, the 48V button. Dynamic microphones do not need it and are not harmed by it, so leaving it on is generally safe.

Gain is how much the interface amplifies the microphone. Too little and you get a quiet recording with audible hiss when you raise it afterwards. Too much and the loud moments clip, which cannot be repaired.

Speak at the loudest you expect to get during the recording, not at a polite test volume, and turn the gain up until the meter peaks around minus 12 to minus 6 dB. That leaves headroom for the moment you get animated.

Most interfaces have a light on each input that glows green at a good level and red when clipping. Aim for green with occasional flickers towards the top, and never red.

Recording a little quiet is recoverable. Recording clipped is not. When uncertain, go lower.

Buffer size sets how much audio the computer processes at once. A small buffer means less delay and more strain on the processor, which shows up as clicks and dropouts. A large buffer is stable and introduces noticeable delay.

For recording a voice with direct hardware monitoring, latency does not matter at all, so set the buffer high, 512 or 1024 samples, and enjoy the stability.

Only lower it if you are monitoring through the software, for instance to hear an effect while you perform. Then work down until you hear glitches and go back one step.

Set the sample rate to 48,000 Hz and leave it. It matches video, it is what most platforms expect, and higher rates make larger files without helping a spoken recording.

If you are recording one person talking, a good USB microphone gives you most of the benefit with none of the setup. It contains its own preamp and converter, it plugs straight in, and there is no phantom power or gain staging to get wrong.

An interface earns its place when you need two or more microphones on separate tracks, when you already own an XLR microphone you like, or when you want to upgrade the microphone later without replacing everything.

Two microphones is the clearest case. A computer can only record from one device at a time, so two USB microphones do not work, and a two-input interface solves that cleanly.

If the money is limited, spend it on the microphone and the room before the interface. Distance to the microphone and soft furnishings change a recording far more than converter quality does.