Learning Notes: Digital Audio Recording
In short, basically: digital audio recording is using an electronic device to record sound.
In short, technically: digital audio recording is translating into binary digits one or more patterns of pressure changes within a longitudinal wave (binary “digits”, hence “digital” recording).
In medium: it’s using a device called an analog to digital converter (ADC (e.g. inside a usb microphone)) to measure two traits of that wave. The mic is designed to physically vibrate when sound waves hit it. When the waves hit it, it creates analog AC voltage. Two traits of that voltage are measured by the ADC: the amplitude (unit deciBels (dB) and the # of times the amplitude pattern repeats per second (called frequency, unit Hertz (Hz)). Amplitude in this context is the maximum (peak) pressure the sound wave produces.
In long: it’s taking a continuous (a.k.a. analog) signal and recording the signal’s amplitude a set # of times per second. That # of times is called sampling rate. Each time it records an amplitude reading, that reading is called a “sample”. That sample amplitude reading is stored as a string of 1s and 0s (i.e. in the form of binary code). The amplitude represented by that binary code is determined by rounding the real (analog) amplitude to the nearest interval allowed by the bit depth setting of the ADC. This rounding process is called “quantization”. The difference between the true amplitude and the amplitude to which it is quantized is called “quantization noise”. Physically, you’re hearing one of two things: either the amplitude of the frequencies you want to hear are being rounded down, or the amplitude of the frequencies you don’t want to hear are being rounded up. Either way, it’s reducing what’s called “signal to quantization noise ratio” (SQNR). Noise, technically, is a non-patterned group of frequencies. Like when you knock on a door. It’s not a note like a piano key, because the frequencies emitted are not patterned as integer harmonies of the main frequency. Like white light, which is all the colors at once, pure white noise is all frequencies at once. That noise is the nasty sound you hear when the audio you hear was recorded at too low of a bit depth. You might call it gritty, or crunchy, or as if there’s just wind coming straight at the mic. Recording at too low of a sample rate is more of a grainy or static-y sound, but it’s still nasty. That’s not quantization, it’s called “aliasing”. Here’s an example of what aliasing and quantization noise sound like: