Sound Energy to Electrical Energy:

When we make an audio recording, we are not storing the actual sound, but rather an electrical representation of it. This is similar to how a photograph stores a representation of a visual scene, rather than the actual subject.

What we perceive as “sound” is actually vibrating air molecules. In order to store this energy, we must convert it from its current form to an electrical signal. This is done using a transducer, which in the case of turning sound energy into electrical energy, is a microphone. Conversely, when turning electrical energy back into sound energy, a speaker is used.

Once the sound energy has been converted to an electrical signal, it can be transferred to a storage medium. It is possible to store multiple signals at once, but this would require the use of multiple channels. An audio channel is an independent signal path for a given signal. People often use the terms “channel” and “track” interchangeably, but they are different entities. A track is a space on a storage medium, such as the different songs on your phone. For example, two audio channels (left and right) carry separate signals to each earbud, creating a stereo effect.

Analog Audio Recording:

Currently, we have two options for storing this information: analog or digital. For most of the twentieth century, analog recording was the sole option. It is based on drawing a complete analogy to the fundamental form of sound energy (waveform). Depending on the storage media (tape, vinyl), several techniques are used to create a copy. Iron particles generate patterns that represent information on a piece of magnetic tape. Grooves carved into vinyl records recreate the waveform. When the needle goes over these grooves, it vibrates just like the original sound waves.

Digital Audio Recording:

In a digital recording, things are a bit different. Rather than creating an exact representation of the signal, the signal is translated into a series of zeroes and ones (binary code). Only this stream of information is actually stored. This is done by a process called sampling. In sampling, small sections of the waveform are taken (like little snapshots) at a very high rate of speed. When these snapshots are played back at the same rate of speed, it sounds continuous to our ears. Think of a cartoon, flip- book, or zoetrope—the static images go by so quickly that your brain is “tricked” into seeing movement. Here, the same concept is used for sound.

The Technical Aspects of Audio Recording

Sample Rate and Sample Depth:

How fast or how often we take these samples is called the sampling or sample rate. Sampling rates are quite fast: 444,100 times per second or 48,000 times per second. One cycle per second is known as a hertz, while 1,000 cycles per second is a kilohertz. So, sample rates are always measured using units of kilohertz. It is important to distinguish sampling rates from frequencies of audible sound. Remember human hearing (at absolute best) is 20 Hz to 20 kHz. Sampling rates are always much higher numbers.

The sampling rate takes care of half of the information we need, however, as there are two sides to the entire equation of recording sound. On one side, we have frequencies, fundamentals, harmonics, and things of that nature; on the other, volume, dynamics, sound pressure level etc. These latter elements are represented through voltage, and the higher the voltage, the louder the sound. We measure this voltage using bits to give us a sample depth or bit resolution.

The recordings that you will be working with will be either 16 or 24 bits. Each bit can be either a one or a zero (on or off) and can store 6 decibels (dB) of dynamic range. These bits can be arranged in myriad combinations, giving us a total of 65,536 possible steps in 16 bit and a total of 96 possible dB of range, and 16,777,216 steps and 144dB for 24 bits. Digital audio is represented on a bar graph: We measure sampling rate from left to right, and bit resolution from top to bottom. The processor that does all of this sampling is called an A to D (analog to digital) convertor.

Dynamic Range:

In order to produce clean recordings, we must be careful about how loud they are. If recordings are made at too high a volume, sound will become distorted. There exists a range of volume information. The term we use to describe the ratio between the smallest and largest signals that can be measured is “dynamic range.” On the lowest side of the range is what is called the “noise floor”—the lowest level any signal can be captured at. The highest side of the range is “overload”—the point at which the signal distorts. For a 16 bit recording, the dynamic range is 96 dB, for 24-bit, 144 dB. If 0 dB represents the point of overload, -96 DB (144 for 24 bit) is the noise floor. A level of -12 dB is our “sweet spot” or reference level. All recorded dialogue should be at this level; anything above it is known as the headroom. Having adequate headroom is important as it gives the recording some room to “breathe” and space to put sounds that are louder than dialogue (think explosion sound effects).

Distortion

Enemy number one in recording audio is distortion, which is defined as any change in the waveform from the original. The most common form of distortion is called clipping and is due to signal overload: The signal is too loud to be captured accurately and the peak of the waveform is cut off.