Wednesday, September 06, 2006

Data and MP3 Compression: Understanding "Digital"

One of the biggest mysteries to many is what data compression is all about and why it matters to me. But before this, you must first understand what digital data is.

The MP3 and MP4 generation was generated mostly due to advances in data compression. Let us first take a step back and focus on digital music. There are two types of data in the world: Analog and Digital data. Analog data is real time data, and it is essentially how we perceive the world. As you listen to any music, you are listening to an analog signal. Everything we see and hear is analog. So what is the significance of digital? Digital is a way of storing and transferring this data. Think of a twelve inch ruler. With this ruler, you can measure any distance between between 0 and 12 inches. Depending on how good you are at reading the ruler, you can probably measure to a resolution of 1/16 of an inch. Even though there is measurable distance between these 1/16 inch marks, the distance is rather insignificant to you. At this point, you can record the distance two ways:

1) You can mark the ruler with a pencil and store it in its true "analog" value
2) You can round the value to the nearest 1/16th and record it. This could be referred to as a "digital" value.

Digital is a collection of zeros and ones that represent a number. Because there are two values, the system is base 2. Our normal system is a base 10 system. For example:

The number 1456 is really:

6x(1) + 5x(10) + 4x(100) + 1x(1000) = 1456

Where the number in parentheses are 10^x power.

Let's try a base 2 number of 100110:

0x(1) + 1x(2) + 1x(4) + 0x(8) + 0x(16) + 1x(32) = 38 (decimal)

Where the number in parentheses are 2^x power.

In this base-2 example, the number is referred to as "6-bits". A typical "byte" is normally "8-bits". So instead of storing the number 38, computers would store the number 100110.

So why would you want to do that? There are 6 digitals in base-2 versus two digitals in base-10. The answer is simple! Ones and zeros are much easier to store (It is like a light switch, either ON or OFF). Further, in a base-2 system, there are only 2 possible solutions. In a base-10 solution, there are 10 solutions per digit. The more solutions you have, the more chance to make errors.

Now let's revisit the the analog signal. In this type of data, there are an infinite number of solutions. Mind you, the error won't be great, but you ARE guaranteed some error every time. Every time you recall a analog signal, the results will be varied to some degree.

Think about your cell phone. About 5 years ago, all cell phones were analog. When you entered areas of poor reception, you received static, distortion and noise. Now, most cells phones are digital. As long as a portion of the ones and zeros are transmitted, you will receive a perfect signal every time. (If you transmit the number 1.2 or 1.5 or 0.8, it will always be read a one! i.e., the signal is immune to noise!) If you cannot even read these simple ones and zeros, you drop off completely. The clarity of the signal is perfect and repeatable, as long as the data is present.

Music works in the same manner. Old-school records and cassette tapes are analog signals. These media will product static and noise and will degrade over time. CDs are digital signals. As long as the compact is readable, the CD player will provide you with the exact sound as it was recorded. The CD itself might degrade over time, but the data will always be stored in its original form.

Since the inception of digitally stored music, the quality of music has remained high and to the same standard as it is today. Compact discs still offer the highest form of music available. The problem with compact discs is that the amount of bits required to store a song is high. That leads us to the modern use of data compression and MP3. (And another article for another day!)

- Matthew Bredel

Visit for iPod and portable media reviews and resources!