Categories
The Digital Age

The 192kHz Lie (Part 1)

Bit depth and sample rate are two terms that can sound a bit scary and technical to most home recording enthusiasts. The purpose of this article is to take some of the mystery out of these industry terms. It is also quite common for the sales guy down at the local Guitar Center to tout the 24/192 capability of this software or soundcard which is why it is clearly superior and you need to buy it now. Some of this is just not important to your typical audio recording. We are going to talk a bit about the theory behind these concepts, take a look at some examples of different settings, and compare the benefits of having higher or lower numbers (higher is not always better). Grab a cup of your favorite recreational drink (mine is water) and let’s get started!

Technical Overview

You probably have a good idea of what a sound wave is. If asked to draw one you could take a pencil and paper, and draw a squiggly line across it. That would be a very good representation of what a sound wave is. It is a very smooth wave that goes from low to high and back, over and over again. The key word here is smooth (Figure 1).

Use Your Bits: Figure 1
Figure 1

Computers and digital technology are very wonderful at a great many things. Smooth is not among them. In the digital realm every piece of data is represented by either a zero or a one, off or on. This does not lend itself to smooth. If you tried to draw that sound wave by using only straight lines that were all the way and all the way down, it wouldn’t look too pretty. It also wouldn’t look much like the nice smooth “snake” you’ve got there now (Figure 2).

Use Your Bits: Figure 2
Figure 2

This is a representation of what we would call 1-bit audio. It is represented by a single bit, or switch, that is either on or off. Audio would sound pretty bad if we had to represent it with a single switch. If we stack multiple bits on top of one another we can get a great resolution, more steps, to represent our wave more accurately in the vertical direction. You’ll notice we have only two steps in our picture (a high and a low). Greater bit depth gives us a greater number of steps, but if we still only have two of them across, we will not get very far. This brings us to the sample rate.

The sample rate is the number of times across the wave horizontally that we get to decide which step it is on. Instead of me rambling on for another thousand words, check out Figure 3.

Use Your Bits: Figure 3
Figure 4

If you were the computer, it would be your job to take that wave and play “connect the dots” by moving across the page, and on each vertical line, place a dot at the horizontal intersection closest to it (Figure 4). The part of the computer that does this is called the analog-to-digital converter (or ADC).

Use Your Bits: Figure 4
Figure 4

We can represent each dot by its coordinates on the graph. This is the data that is stored as digital data (Figure 5).

Use Your Bits: Figure 5
Figure 5

Still with me? Good. These number coordinates are all stored but eventually we need to be able to hear them again or they are not doing us any good. The job of turning these numbers back into a sound wave falls on the digital-to-analog converter (or DAC). A wave is not a bunch of dots but a continuous fluctuating wave of air pressure. The DAC has to play connect the dots to turn this back into a wave. Hopefully you have played connect the dots enough to be able to envision what this ends up looking like. Looking at Figure 5 it is not hard to imagine it looks pretty close to our original wave. Check out what happens if we do a direct comparison though (Figure 6).

Use Your Bits: Figure 6
Figure 6

It is close to what we put in but not exact. Notice there are some places where we are quite a bit off from our original analog wave. Can you think of a way to get our digital representation closer to the original? Remember, we can only play connect the dots on intersections of lines. Figure it out yet? You are correct, we get close by increasing the number of lines. We have three choices for how to increase the number of lines (and therefore intersections):

  • put in more vertical lines (higher sample rate)
  • put in more horizontal lines (greater bit depth)
  • both (greater bit depth and higher sample rate)

We are going to explore the cost and benefit of all three so you can decide which of the three solutions is best for you.

Continued in The 192kHz Lie (Part 2)

5 replies on “The 192kHz Lie (Part 1)”

I like your website as well as this article. For the longest time all my mixes were done at 16-bit 44khz. They were “good” but not the greatest. My pc could never handle anything higher. But thanks to low cost technology these days, I easily run 24-bit 44khz. I can agree that you don’t need anything higher than that. I cannot really imagine why anyone would want to?? Is this just really some kind of joke put out there by pro studios? I mean they use to get away with saying that if it aint Pro-Tools then it aint PRO. I’ve been using Cakewalk since the beginning and it will do commercial quality as well as any. After all, it’s the ears that do the creating. Not the wallet.

Dear Ben,

I was drawn to your site through a link in the Jamstix website. I respect and appreciate your work here. However, I did feel you are presenting misleading information about sampling with this article. Your description of sampling is the common misconception about the interpretation and construction of digital signals. This is not the case! Digital sampling recreates EXACTLY all the information at half the frequency of the sampling rate and below, i.e. at 44.1 kHz we get perfect reproduction of all signals below 22 kHz. There are artifacts that are created when recording signals higher than 22 kHz, but most hardware has filtering to correct for that. A good (and readable) essay on the details can be found at http://www.lavryengineering.com/documents/Sampling_Theory.pdf . This is very complete but the first 3 pages go over the same point you are making.

This is FYI and not meant to be any form of criticism. I hope you find this usefull if not already understood.

Yours Truly,

Chris Brown

Indeed, I agree with what you’re saying somewhat. Don’t forget that while this is true theoretically, if it were true practically then why aren’t we saving tons of space by recording everything at 4-bits resolution? There is a huge gain to be had in the capturing of low level sounds like reverb tails when you move from 16-bit to 24-bit. The gains in upper harmonics when you go from 44.1k to 192k are much more subjective.

Comments are closed.