Warning: This information verges on geek-speak, but it will help you know what you’re doing when you set up your digital recording system.

 


 

A few years ago, all music was recorded on analog tape recorders. They record the audio signal as magnetic patterns that rise and fall in strength as the signal waves do. Digital recorders, on the other hand, store the audio signal as a numerical code of ones and zeros.

 

Analog tape recordings have a little tape hiss, distortion and unsteady pitch. In contrast, digital recordings don’t have those problems, so they sound very clean. Compared to analog tape recorders, digital recorders tend to be smaller and lower cost. They can record timing information. Also, they have random-access: they can go to a particular section of the recording very quickly.

 


 

Current digital recorders come in many formats:

-Hard-drive recorder

-MiniDisc recorder

-CD recorder

-Computer recorder (software that lets you record on the hard drive)

-Flash-memory recorder

-Mixer with an internal hard-drive recorder

 

Digital Recording

Like an analog recorder, a digital recorder puts audio on a recording medium, but in a different way. Here’s what happens in the most common digital recording method, called Pulse Code Modulation or PCM:

 

 

 

Figure 1. Analog-to-digital conversion (during recording)

 


1) The signal from your mixer (Figure 1-A) is a varying voltage. This signal is run through a lowpass filter (anti-aliasing filter) which removes all frequencies above 20 kHz.

2) Next, the filtered signal passes through an analog-to-digital (A/D) converter. This converter measures the changing voltage of the signal several thousand times a second (Figure 1-B).

3) Each time the waveform is measured, a binary number (made of 1’s and 0’s) is generated. This number is the voltage of the signal at the instant it is measured (Figure 1-C). Each 1 and 0 is called a bit, which stands for binary digit.

4) Those binary numbers are stored on the recording medium (Figure 1-D). The numbers can be stored on tape, hard disk, compact disc, MiniDisc or flash-memory card.

 

 

Figure 2. Digital-to-analog conversion (during playback)

 

 

The playback process is the reverse: The binary numbers are read from the recording medium (Figure 2-A) The digital-to-analog (D/ A) converter translates the numbers back into an analog signal made of voltage steps (Figure 2-B).

 

An anti-imaging filter (lowpass filter) smooths the steps in the analog signal (Figure 2-C), and the smoothed signal leaves the D/A converter. The original signal’s waveform is reproduced.

 

 

 

Figure 3. An analogy of digital recording and playback: audio, recreating a disk shape using wooden strips

 

 

As an analogy, suppose you want to reproduce the circular shape of a wooden disk.

1) You lay a ruler across the disk every 1/4 inch and measure how wide the disk is at each point (Figure 3-A).

2) You cut a bunch of wood strips with the same lengths as those measurements, and put the strips side-by-side. They look like stair steps going up and down (Figure 3-B).

 

You file the raggedy edges of the wood strips so they join together smoothly. That’s like the anti-imaging filter. You’ve just reproduced the shape of the original wooden disk (Figure 3- C).

 

Compared to analog recording, digital recording has inaudible noise, distortion, and speed variations. That’s because the hard-drive head reads only 1’s and 0’s, so it is insensitive to the magnetic disk’s noise and distortion. During recording and playback, numbers are read into a buffer memory and read out at a constant rate, eliminating speed variations in the hard disk. Special coding during recording, and decoding during playback, corrects for missing bits by using redundant data.

 

Digital audio is usually recorded as a wave file or AIFF file. Both are standard formats for audio files. Wave (.wav) is for PC; AIFF (Audio Interchange File Format) is for Mac. Both formats use “linear PCM encoding” which has no data compression (unlike MP3 or WMA formats).

 

Bit Depth

As we said, the audio signal is measured many thousand times a second. Each measurement generates a group of binary numbers called a “word”. The longer each word is (the more bits it has), the greater is the accuracy of each measurement. Short words (8 bits or less) give poor resolution of the signal voltage, causing high distortion and high noise. Long words (16 bits or higher) give good resolution, which results in low distortion and low noise. Bit depth or resolution are other terms for word length.

 

In our wooden disk analogy, bit depth (word length) corresponds to the accuracy of the ruler measurement: whether we measured to the closest 1/8”, 1/16”, 1/32”, etc.

 

A word length of 16 bits is adequate (but not optimum) for hi-fi reproduction. It is the current standard for compact disc. Most digital recorders offer 24-bit resolution, which sounds a little smoother and more transparent than 16-bit, but needs more disk storage space.

 

Even though CDs are a 16-bit format, they sound better when made from 24-bit recordings. During mastering, you add dither (low-level noise) to the 24-bit recording, then export or save it as a 16-bit recording. You copy the 16-bit recording to CD. The dither helps the 16-bit recording sound more like the 24-bit recording. In other words, dither lets you retain most of the quality and resolution that you recorded at 24 bits, even though the recording ends up on a 16-bit CD.

 

Sampling Rate

Sampling rate is the rate at which the A-to-D converter measures (samples) the analog signal while recording. For example, a rate of 48 kHz is 48,000 measurements per second. That is, 48,000 measurements are generated for each second of sound. The higher the sampling rate, the higher the frequency response of the recording. The highest frequency you can record is one-half the sampling rate. Compact discs use a 44.1 kHz sampling rate and so their frequency response extends to 22.05 kHz, a little beyond the upper range of human hearing.

 

In our wooden disk analogy, sampling rate corresponds to the number of wood strips you cut to form the disk. If you measure the disk every 1/2” across, that is like using a low sampling rate. If you measure the disk every 1/8” across, that is like using a high sampling rate.

 

Sampling rates for high-quality audio can be 44.1 kHz, 48 kHz, 88 kHz, 96 kHz, or 192 kHz. The higher sampling rates sound smoother and more transparent but need more disk storage space. CDs are recorded at 44.1 kHz sampling rate, which many engineers consider to be adequate. A 96 kHz sampling rate can be used in DVD-Audio discs.

 

In summary, a digital recorder measures the analog signal several thousand times a second, and generates a string of 1’s and 0’s that represents those measurements. Sampling rate affects the high-frequency response. Bit depth affects the dynamic range, noise, and distortion. The sound quality of any digital recorder depends mainly on the accuracy of its A/D and D/A converters.

 

Digital Recording Level

In a digital recorder, the r
ecord-level meter is a peak-reading bargraph meter that reads up to 0 dBFS (FS means full scale). In a 16-bit digital recording, 0 dBFS means all 16 bits are on. In a 24-bit digital recording, 0 dBFS means that all 24 bits are on. When you set the recording level, it’s a good idea to aim for -6 dB to -3 dB maximum peak level so that unexpected peaks don’t exceed 0 dBFS, which causes ugly-sounding clipping distortion. If you’re making a 24-bit recording, the recording level is not very critical because a 16-bit signal is at –48 dBFS!

 

We’ve scratched the surface of digital audio, and there’s always more to learn.