The hook
You've heard people argue about it in the lobby: analog is warmer, digital is cleaner, vinyl is truth, whatever. Meanwhile you're standing in front of a digital console wondering something much more basic that nobody will just answer: when the singer sings, what is actually traveling down that cable? Is it... sound? Tiny sound?
It's not sound. And once you know what it really is, half the mystery of the whole system evaporates.
The lesson
Follow one word from the singer's mouth into the board.
The singer makes a pressure wave (lesson 1.1). The wave hits the microphone, and the mic does one job: it converts the pressure wiggle into an identical electrical wiggle. Pressure goes up, voltage goes up. Pressure falls, voltage falls. The voltage on the cable is a copy of the sound wave, drawn in electricity instead of air. An analog of it. That's where the word comes from.
That voltage is genuinely tiny: a mic puts out a few thousandths of a volt. It gets boosted by the preamp (lesson 0.1's gain knob), and in an all-analog system it stays a voltage the entire trip: through the EQ, down the faders, out to the amplifier, which enlarges it enough to shove a speaker cone back and forth and turn it back into a pressure wave. Air to voltage to bigger voltage to air again. That's the whole magic trick of audio, and until the 1990s that's all there was.
A digital console adds one step, and it happens at the very front door. Right after the preamp, an analog-to-digital converter measures the voltage (takes a snapshot of it) thousands of times per second, and writes each snapshot down as a number. From that point on, "the signal" is a stream of numbers, and everything the console does to it (EQ, compression, mixing, routing) is arithmetic. At the output, a digital-to-analog converter turns the numbers back into a voltage for the amplifier.
Two numbers describe how good those snapshots are, and they're the two specs you see everywhere:
Sample rate: how often we snapshot. Live and broadcast audio runs at 48,000 snapshots per second (48 kHz). Why so many? There's a hard mathematical rule: to capture a frequency, you need at least two samples per cycle. So 48 kHz can faithfully capture everything up to 24 kHz, comfortably past the 20 kHz edge of human hearing. Everything you can hear fits, with margin.
Bit depth: how precisely each snapshot is measured. Each snapshot's voltage gets rounded to the nearest step on a ruler. More bits = a ruler with finer steps. Every extra bit adds about 6 dB of range between the quietest and loudest thing the system can represent. Modern consoles measure with 24 bits: a ruler with over sixteen million steps, good for around 144 dB of theoretical range. That's wider than the range from a pin drop to a jet engine, and comfortably wider than your ears (lesson 1.3) or any PA.
Here's what those numbers mean in practice: a modern digital console captures audio more accurately than any analog console ever did. The "digital sounds cold" arguments date from early converters in the 1980s that genuinely were rough. That war is over. Today you choose digital or analog for workflow reasons (recall, scenes, flexibility, price), not fidelity.
But digital does change two things you must respect at the board:
The ceiling is absolute. An analog console pushed too hard distorts gradually. The transformers and circuits saturate, and a little of it can even sound pleasant. A digital system has a hard numeric maximum: 0 dBFS, from lesson 1.2. Push past it and the waveform doesn't bend, it flattens: the converter just writes the maximum number over and over. The result is harsh and useless. There is no "a little over" in digital. This is why gain staging (coming in Level 2) matters more, not less, on modern boards.
Everything takes a moment. Converting, computing, and converting back takes time: a few milliseconds through a typical digital console and PA. It's called latency. A few milliseconds is nothing for the congregation (sound already takes about 9 ms to travel 10 feet through air), but it becomes a real issue for in-ear monitors, and a starring problem in the livestream lessons. File it away for now.
Go deeperoptional
The two-samples-per-cycle rule is the Nyquist–Shannon sampling theorem: a signal containing no energy above frequency f can be perfectly reconstructed from samples taken at a rate of at least 2f. Not approximately — perfectly, in the mathematical sense. The stair-step picture of digital audio is a myth; reconstruction (sinc interpolation in theory, the D/A's reconstruction filter in practice) recovers the original smooth waveform exactly, provided the input was properly band-limited.
That proviso is where the engineering lives. Any energy above the Nyquist frequency (f_s/2) doesn't disappear. It aliases: it folds back into the audible band as inharmonic garbage that no filter can remove afterward. So every A/D runs an anti-aliasing low-pass filter in front of the converter. At 44.1/48 kHz that filter must fall off a cliff between 20 kHz and Nyquist, and steep filters ring; this is the legitimate technical argument for 96 kHz — not "more highs," but a gentler filter with better phase behavior. Modern oversampling converters largely dissolve the issue by sampling internally at MHz rates and doing the steep filtering in the digital domain.
Bit depth: quantizing to N bits yields a signal-to-quantization-noise ratio of 6.02·N + 1.76 dB: 98 dB at 16-bit, 146 dB at 24-bit (real converters deliver 110–125 dB; analog stages, not math, set the floor). Quantization error would be a distortion correlated with the signal, but dither (a tiny amount of noise deliberately added before rounding) decorrelates it and turns the distortion into a benign, steady noise floor. It's one of the great counterintuitive results in audio: adding noise increases fidelity.
Inside the console, the mixing math runs at higher precision still (32-bit float or 40-bit-plus fixed point), so internal buses have effectively unlimited headroom; the hard 0 dBFS ceilings live at the converters, on the way in and the way out. That's why clipping an input gain is fatal but running an internal bus "hot" often isn't.
Latency arithmetic worth carrying: at 48 kHz, one sample is 20.8 µs. Converter pairs cost roughly 0.5–1.5 ms round trip; console DSP a fraction of a ms to a few ms; a full digital chain (console + processor + amps) typically lands in the 2–8 ms region. Compare with air: sound travels at about 1 foot per ms, so a monitor wedge 6 feet away "adds" 6 ms all by itself. Digital latency only becomes perceptually dangerous for in-ears, where the path from mouth to ear is otherwise near-zero. That story continues in Level 6.
Remember
- The signal is a voltage copying the sound wave's shape. A mic converts air to voltage; a speaker converts it back.
- Digital consoles measure that voltage 48,000 times a second (sample rate) with 24-bit precision (bit depth), then do everything as math.
- 48 kHz captures everything humans can hear, with margin. 24-bit gives more dynamic range than any real-world stage. Modern conversion is not the weak link.
- Digital's ceiling is absolute: 0 dBFS. Analog fades into distortion; digital slams into a wall. Respect the gain.
- Digital adds a few milliseconds of latency: irrelevant in the room, important later for in-ears and streams.
Your turn
- Trace the journey out loud: pressure wave → mic → tiny voltage → preamp → converter → numbers → converter → big voltage → speaker cone → pressure wave. No notes.
- Find your console's sample rate in its setup menu.
- Explain to someone why there's no such thing as "slightly clipped" in digital.
- Rough math: your board runs 48 kHz at 24-bit on 32 inputs. About how many numbers per second is it handling from the stage? (Ballpark is the point.)