So, how do analog and digital differ from one another? Think about an analog sine wave. It looks like a nice, smooth line (See chart, this page.) Now if you take that same analog sine wave, digitize it and look very closely, it is not so smooth anymore - as indicated by the stair-step "curve" in the same illustration.

This will be a two-part article. The first part will go over what digital audio is made up of. The second part will go over some ways of how to get the best out of your digital sound devices.

Samples, Bits and Pieces

There's obviously a lot to digital audio. That's why people out there are still trying to replicate gear that was made over 50 years ago! There are two basic components of what digital audio is made up of, bit depth (also known as length) and sample rate.

In the pro audio world, bit depth is most commonly seen as being 16-bit or 24-bit. This does not refer to frequency response, but can be described as to how many "steps" in volume a digitized signal can possibly have. The actual number refers to a binary "word length," and each binary digit within that word will have a value of either a "0" or a "1". The combination of these 0s and 1s is what makes up the value of a "bit" in the digital realm.

As an example, a 1-bit word would be "1" digit in length, having 2 steps in volume, either "on" or "off." Conversely, a 16-bit "word length" would be 16 digits in length, having 96 steps in volume, and 24-bit having 144. The rule of thumb is that every "1 bit" will give six more steps in volume. Analog, on the other hand, can be described as having an infinite word length, because the signal is continuous.

A "sample rate" can be defined by looking at exactly what the two words mean. A "sample" is a single record of data that contains everything about the audio signal at that present point in time; level (volume), phase, and the combining amplitude of frequencies (creating the frequency response). The "rate" is commonly seen as 44.1k, 48k, 96k, etc. The "k" represents that the number it is next to is multiplied by "1,000." For example, 44.1k stands for 44,100 and 48k stands for 48,000. This number is the "rate" at which these samples are recorded/played back. For example if a digital console is processing a signal at "48k," that means the signal is "sampled" 48,000 times per second while in the digital realm of that console.

The sampling rate at which a given device operates will determine what bandwidth that device will need to have. This is where the "Nyquist Theorem" comes into play - to reproduce any given frequency, you must sample at least twice that frequency. For example, to reproduce 20,000 Hz, you must sample at least 40,000 samples per second. It can and will be argued that one should "oversample" a signal to obtain a higher "sonic" resolution of the transient sounds in a signal. That is why we sometimes see 96k and even 192k. Think about that basic analog sine wave again. Now digitize it and think about what it would look like with a 44.1k sampling rate, and then with a 192k sampling rate. The 192k sampling rate would much more resemble the analog sound were trying to replicate, correct? Exactly!

How Do We Get Here?

The conversion process of audio signals can be described as being one of the arch nemeses of the digital audio world. This is where the analog signal is converted into a digital signal (A-D), and when the digital signal is converted back into an analog signal (D-A).

There are many digital devices out now on the market; digital consoles, snakes, equalizers, crossovers, amps and playback units (just to name a few). Some of these devices have digital inputs and outputs alongside the analog ones. Every time an analog signal goes into an input or out of an output, there is a conversion (the A-D and D-A). These digital inputs and outputs are put in place to help keep the signal in the "digital realm" while going through multiple devices to keep converting to a minimum.

Some of the more popular digital connections are S/PDIF, AES/EBU, Lightpipe and MADI, all of which operate in a unidirectional manner. This means one cable will carry the various channels of audio in one direction down a cable, requiring two cables if a send and receive is needed for the same device. S/PDIF and AES/EBU operate in a similar manner carrying two channels of audio, but differ in a couple ways, including connector types and ohm resistance of the cable required. Lightpipe sends audio via an optical cable and contains up to eight channels of audio. The current version of MADI uses a BNC connector with a 75-ohm coaxial cable and carries up to 64 channels. The older one still uses a BNC connector, but only carries 56 channels of audio.

So now that we have gone over what digital audio is made up of, you can see there is a plethora of selection between device and connection types. Next month we will go over some of the better ways to interchange these different devices and get the best out of your digital gear.