I finally learn the true meaning of TLDR - I can only remember Thomas Edison from that episode of Drunk History but he's a jerk, right? - Wax cylinders: nature's mp3 - I make audio way more confusing by introducing a robot analogy - I bring things back to solid ground...with another robot analogy
I'm starting with the helpful tip this time. TLDR: always use 24 bits when recording and processing audio. This is one of the least controversial points of working with digital audio.
There, I did it friends. I got to the point.
Now for the rest of it.
A Bit of History (but not too much)
Remember when I talked about how amazing it is that an entire Taylor Swift song can be represented as a single waveform? You probably don't. I don't actually think I used Taylor Swift in that example anyway. Well, it still blows my mind every time I think about it. Your little cellphone speaker can play a single, mono waveform and you will hear drums, guitars, cymbals, bass guitars, pianos, Taylor's voice and whatever else goes into a Taylor Swift production.
When recorded audio was first invented, they used used this principle in the simplest way possible. The year was 1877. That lovable rascal Tommy Edison and his mischievous group of ragamuffins had just invented the phonograph - a predecessor to the record player. The phonograph engraved a sound waveform onto a rotating cylinder and was then able to play it back. The needle followed the grooves in the cylinder and, as it did, created vibrations that matched (sort of) and recreated the original waveform. Amazingly, you can hear one of the earliest recordings ever made using a phonograph. Check it out here
Analog Signals - A Robot on a Trampoline
Records eventually replacing wax cylinders (ever wonder how they get stereo sound on a record?).
Then we had magnetic tape. All of these methods are similar in that they're capturing and reproducing audio in the analog domain. This means that there is some continuous waveform (think about the grooves cut in the record) and the recording process captures and reproduces this waveform. Let's get into some visuals.
Pretend we have a robot (pronounced: RO-bit) and this robot is jumping on a trampoline (just bear with me).
This robot represents the amplitude of our waveform. In other words, if the robot jumps very high, that represents a loud signal. If the robot jumps very lightly, that represents a quiet signal. If we graphed the robots position over time, we would get a waveform. We can then play that waveform back to recreate the original "event".
I know right now it seems like I've just taken a simple concept and made it WAY more complicated than it needs to be. You'll see where this is going in a minute...I think.
The process we just described is an analog process. The robot's position in the air does not have discrete positions. It can move in a smooth trajectory through space.
Digital Signals - Replace the Trampoline with a Ladder
Digital processes are not "smooth". Digital processes are made up of little discrete steps. This is because the traditional computers we all use (based on Alan Touring's ideas) operate on the principle of digital bits. A bit is a little switch that can be either on or off (1 or 0).
What does this mean for our robot? Well instead of moving smoothly through space as it jumps on the trampoline, we can think of the robot as using a ladder to move up and down. The ladder is different because it has discrete steps. The robot can occupy each step but cannot really "occupy" the space between steps.
Now, the more steps the ladder has, the smoother the robot's path up and down will look. If there are only 5 rungs, the robot will be taking big, noticeable steps (like in the annoying animation above).
Side note: I hate when websites have unstoppable GIFs because I have cat-like ADD and get easily distracted. Sorry friends).
If there are 400 rungs, the robot will take tiny little steps and, because he's an excellent climber, his trajectory will start to resemble what it was on the trampoline. If there are 40,000 little ladder rungs, we may not even be able to tell the difference between the ladder process and the trampoline process.
So, what is bit depth in digital audio? Bit depth is the number of rungs on the ladder.
At 16 bits (the standard for CDs - remember CDs?) the ladder has 2^16 = 65,536 rungs.
At 24 bits (what you should be recording at), the ladder has 16,777,216 rungs. A big difference, right?
More than 24 is not necessary, but less is not recommended - for reasons which I'll get into in some future post.
Maybe.
Happy recording!
Vadim
Comments