2. What is Digital Audio?
It would be difficult for anyone who reads the news or watches TV to avoid the constant message that the world is going digital, and that digital is better. Digital audio is better than other audio, digital video is better than other video, etc. If you are like a lot of us, you are wondering just what digital audio is, and whether it really is better. The following information should give you a clearer idea of what digital audio is and what it is good for.
What is sound?
The dictionary defines "audio" as "audible sound reproduced mechanically". Like so many dictionary definitions, this leads to a new question: what, exactly, is sound? We all know it when we hear it, but what is it?
In its simplest terms sound is just a vibration that is transmitted through the air to our ears. Most of us have put our hand on a mechanical device that is vibrating and felt the vibration. Normally we also hear the vibration as it affects the air around us. Sound "waves" are successive areas of air compression or rarefaction. The speed of sound is simply the speed at which those areas of compression and rarefaction pass through the atmosphere.
Think of what happens when you hit a drum. The air directly under the drumhead is compressed. Next to the area of compression there must be an area with less air- an area of rarefication. This is caused by the drumhead bouncing back up. In fact, the drumhead will vibrate back and forth several times, creating a series of areas that are compressed next to areas that are rarefied.
The areas of compressed and rarefied air move out from the drumhead just like ripples on a pond. We call this a sound wave. When it gets to your ears your eardrums move to match the air pressure, and nerves inside your ear pick up the movement and send it to your brain as sound.
Looking at Sound
If you have ever used a wave editor, or watched someone else use one, you have probably seen a graph of the level of air compression or rarefaction.
If you havenít, here is a waveform created by Blaze Audio's RipEditBurn Plus.
(If you want to experiment with sound waves you can download a free trial version of RipEditBurn Plus.)
Usually there is a horizontal line in the middle of the display representing no compression or rarefaction. The line representing the sound goes up as air becomes more compressed and down as it becomes less compressed. These changes happen very quickly. The graph of the compression/rarefaction changes over time is often called the "waveform". Experienced audio engineers can often tell quite a bit about how a recording will sound by viewing the waveform.
Why would someone who is not an audio engineer want to view a waveform? Suppose you made a recording, but there is a loud noise in the middle. With a sound editor and a little experimentation, you can find and even remove that noise! In the image below, in the highlighted area you can see the click on a track recorded from an lp.
In the real world the level of air compression or rarefaction changes smoothly. Even a very quick change in the pressure of the air is still a smooth change from one point to another. Analog recordings (usually done today on tape) store all of these smooth changes. The amount of magnetic energy stored on the tape moves smoothly up and down as the intensity of the sound moves up and down.
Computers and other digital equipment are not designed to handle these continuous and gradual changes. Instead, they only understand two values- on and off. The magnetic energy stored on a computer tape or disk consists of ones and zeros- nothing else. If you fed a computer tape directly through an amplifier into a speaker the speaker would interpret the ones as full power and the zeros as no power. The result, instead of being a smooth, gradual shift from one value to another, would be an ugly buzz as the speaker cone tried to keep up with these abrupt changes.
(Note- playing digital data through your stereo system or computer speakers can be hard on your speakers. Take our word for what it sounds like, or turn the volume way down!)
Fortunately thatís not what happens when playing digital audio. First of all, the ones and zeros are grouped together (normally 8, 16, 24, or 32 at a time) to form larger numbers in the binary numbering system. For example, 00001001 in binary would be read by the computer as the decimal number 9. The sequence of numbers:
translates into 9 and 7 in decimal. In this example, using 16 bit numbers, the minimum value would be 0 and the maximum value would be 65335. The change from 9 to 7 within that range would be quite gradual, although not as smooth as would be experienced in an analog system.
Digital audio takes advantage of some peculiarities of acoustics and the human ear. An analog waveform would contain every value between 9 and 7 for at least a very brief length of time. No computer could hold all of these numbers. So, when sound is converted from analog into digital audio, the hardware "samples" the level of the waveform at a specific interval. For CD audio, this interval is 1/44,100th of a second. In other words, 44,100 times each second a special chip calculates a value for analog input and sends it off for use or storage. This process is called "digitizing" a sound.
The result, if we were to graph it as we have with analog waveforms, would look quite different. Instead of smooth, gradual changes we would see stair steps as the line jerked from sampling data point to sampling data point. Hereís a picture showing the difference:
There are two useful terms here- the "sampling rate" and the "sample size". The sample rate is the number of times per second that the analog signal is measured. The sample size tells us what number is associated with the maximum value. The maximum value of the analog data doesnít change- if you try to add power past a certain point you just start blowing up hardware. But if the range is from 0 to 1000 the values stored will represent the analog data more closely than if they range from 0 to 10.
Still, if analog data is smooth and digital data is made up of stair steps, why doesnít digital audio sound bad? The answer is fairly technical, but what it boils down to is that, as long as the samples are taken often enough, the noise created by the stair stepping is too high in frequency for us to hear. According to the theory, the frequency of this noise will always be at least twice the sampling frequency. This is called the Nyquist Limit.
Very few if any humans can hear above about 20,000 cycles per second. Note that the speed chosen for audio CDs is 44,100 cycles per second. It is no coincidence that CD sampling rate is just over twice what our ears can hear.
Making Analog Recordings into Digital Files
The conversion of analog data to digital and back to analog is accomplished by special chips. A chip that converts analog to digital is called an ADC- an Analog to Digital Converter. The ADC measures the amount of current at each sampling interval and converts it to a binary number. This is called "digitizing" the sound. On the other end is a chip called a DAC- a Digital to Analog Converter. This chip takes a binary number and converts it to an output voltage.
Here's what happens if you record your voice using a microphone plugged into your computer, then edit it and play it back over your speaker system: the microphone generates an analog waveform corresponding to the compression and rarefaction cycles generated by your voice. This smooth analog waveform is converted into a series of binary values by the ADC which are then transferred into the memory of your computer. Once you are done editing (if youíve ever tried editing analog tape youíll appreciate how much easier digital editing is!) the computer sends the resulting series of binary numbers to the DAC, which converts them to a (relatively) smooth analog waveform that drives your speaker.
Is digital better? Think of the seconds readout of a digital clock as opposed to a smooth sweep second hand on an analog clock. The analog clock can be more accurate, since it shows all of the positions between seconds. The digital clock can be more precise, since it only shows the exact second. Each approach has benefits and drawbacks.
Advantages of Digitizing Audio
Digital, or digitized, sound is easier to reproduce and manipulate without loss in quality. Some question whether the quality is quite as good as analog sound, but it can be very good indeed, and CDs donít wear out like records used to. Digital audio can also be compressed much more easily than analog, which is why MP3 is a digital format.
Digital is not necessarily better, but it is different, and offers advantages to engineers and end users that will increase its dominance in the coming years.