What is the difference between digital recording and analog recording, you may be asking. The answer is both simple… and complex.
At the simplest level, analog recorders－such as tape recorders－and analog playback systems－such as LPs and cassettes－all use a signal physically embedded in a medium. In a tape recorder, that medium is magnetized particles of iron oxide. As the magnetized particles go past the recorder’s tape head, their magnetic state is picked up by coils in the tape head. The faster the tape speed, and the wider the tape, the finer the detail the head can pick up.
On a vinyl record the medium is grooves on a PVC disc. The needle (stylus or pickup) negotiates about 1,500 feet of groove at a speed starting at 20 inches per second (1.13 mph) at the rim, and ending at 8.3 ips as the needle nears the label (0.5 mph). In 40 minutes－the average length of the average LP (both sides)－your stylus travels about two-thirds of a mile. As with tape, the greater the speed the more high frequency information can be embedded, hence the fad of 45 rpm records a few years ago.
Digital recording, by contrast, does not record the signal at all. It first encodes the signal, then records the code. On playback the code is decoded, reconstructing the original audio signal. At no point does the medium ever enter into the signal path, as only the code is retrieved and decoded.
Because of this, digital signals can be copied, stored or transmitted without any loss of fidelity. As long as the code is readable, the output will exactly match the input.
The accuracy of the code, like analog recording, is dependent on the speed (and depth) of the coding. For Compact Discs the speed is 44,100 samples per second, at a depth of 16 bits. Higher speeds give theoretically higher resolution, although whether the difference is audible to human ears is debatable. Blind listening tests usually result in clear proof that you can’t.
The depth of the coding, 16 bits or 24 bits (or more), is how many different values the individual sample could have. The bit “depth” is the power of two that it represents; 16-bit is 2 to the 16th power, or 65,536 possible volumes for that one bit. A 24-bit has over 16 million possible values. As you can imagine, the difference is subtle… to the extreme.
A common question about digital recording is “what happens to the sound between the bits?” The answer to that gets to the heart of digital recording.
At 44.1 kbps (kilobits per second), each millisecond (1/1000th of a second) is represented by 44.1 bits. Zoom in to 1/1000th of a second on a file and it looks like this:
As you can see, there are 44 samples representing just this 1/1000th of a second… and there are a thousand of these per second! That’s about 13 million samples in a five minute song… each of which has 65 thousand possible values. Needless to say this level of resolution is well beyond the ability of humans to discern. The Nyquist theorem (look it up) dictates that the digital sampling frequency needs to be twice the wave frequency to capture a waveform without distortion. Therefore the CD standard captures up to 22kHz, which is beyond even the best human hearing.
On the other end of the frequency scale, there is no limit to how low digital recording can capture. It is theoretically possible to capture 0 Hz, although our ears are generally limited to 20 Hz. Below that, sounds are FELT rather than HEARD… if your stereo reproduces it. (Hint: most don’t.)
So what’s “between the samples”? Well by definition, anything that falls between the samples is higher than 22,000 cycles, so you cannot hear it (a passing bat might). Some audiophiles claim to be able to hear a difference, but again in blind listening tests, they routinely fail to correctly identify the higher sampling rate.
In fact it is provable, mathematically, that one waveform－and only one waveform－will result from a given set of samples. Since that same waveform also CREATED the samples －and the samples haven’t changed－the playback is guaranteed to match the input waveform－exactly.
One additional comment I’ve heard since I wrote this. Some people will try to tell you that higher sampling rates (48k, 96k or even 128k) give you “better resolution.” On the surface that makes sense, right? More bits = greater resolution, just like the analogue formats. But here’s the science. Those additional bits only encode HIGHER frequencies (up to 24k, 48k, 64k) per the Nyquist-Shannon theorem－not additional detail and resolution in the lower frequencies. Your waveform below 22kHz will still be EXACTLY the same waveform, regardless of whether you use 44.1k, 48k, 96k or 128k. What’s “between the bits” is not additional detail… it’s higher frequencies. Frequencies the human ear cannot hear. And most equipment will not play.