The world of audio - and more specifically the realm of audio processing - has its own lexicon of terms that may not be familiar to all users. Many of these words are used on a recurring basis throughout this user manual. To untangle the slang, we've provided a short glossary of terms below.
Note - A quick sidebar for our fellow audio processing gurus: We realize that some of these explanations are rather broad and that we are taking a bit of liberty here and there. Our goal is to strike a balance between "too simple to be helpful" and "my head hurts from all this technobabble," not provide a deep and nuanced technical explanation (as much as blathering on about such things does give us a certain guilty pleasure).
AGC – An acronym for Automatic Gain Control. In processing, an AGC is a type of compression that evens out audio levels by increasing the gain of quieter material and decreasing the gain of louder material. An AGC attempts to keep the output audio at or near a pre-determined target level no matter how much the levels vary at its input.
Attack Time – A setting on a compressor or limiter that determines how quickly it reacts in order to lower the level of an incoming signal when it exceeds a certain threshold.
Brilliance – The highest frequencies of the audio spectrum (roughly those above 8 kHz) that add a sense of “sparkle” or “airiness” to the sound. If over-boosted, it can make the audio irritatingly bright and exaggerate unwanted noise and codec-induced artifacts.
Clipping – A form of waveform distortion that is generally to be avoided as it “clips” the tops of audio peaks and results in unpleasant audible distortion, particularly in digital audio. When properly managed, however, the “distortion” can be used creatively to create a desirable effect (such as in a bass clipper). In AM and FM broadcast applications, clipping of some kind is routinely used to build loudness and as a means of absolute peak control.
Codec – In audio terms, a codec encodes a data stream at its source (either as an audio file or an audio stream) and decodes it at its destination (in a media player or web browser). The most common codes used for audio storage and transmission are MP3 and AAC. Both are “lossy” codecs, meaning they use a data compression algorithm to remove bits of data to reduce the size of a source file or the bandwidth of the audio stream. Higher bitrates throw away less data to try and preserve as much audio quality as possible, while lower bitrates throw away more data at the expense of audio quality in order to minimize file size or bandwidth requirements.
Compression (Audio) - The process of reducing dynamic range by raising the gain of low-level audio and lowering the gain of high level-audio. In the example below, the top waveform is the original unprocessed audio. The bottom waveform has been passed through a compressor.
Compression (Data) - The process of reducing the number of bits in a digital file in order to reduce the amount of required storage capacity or the amount of bandwidth needed to deliver a data stream (including an audio stream). See Codec.
Drive – The amount of gain introduced at the input of an AGC, compressor, or limiter. Depending on the design of a particular circuit, more drive generally results in more overall processing and an increase in output level.
Density – A term used to describe how “processed” audio sounds, specifically in terms of how much the dynamic range has been reduced. Audio with less dynamic range is often described as sounding “compressed” or “dense,” as compared to audio with more dynamic range which is often described as sounding “open.” On the plus side, more density can make an audio stream more intelligible and easier to listen to in noisy environments or on small speakers or earbuds. Too much density, however, can result in a flat sound that robs music of the natural variations in levels that add to its enjoyment and ultimately results in listener fatigue and tune-out.
Dynamic Range – The ratio between the softest and loudest sounds on a recording. Technically, and in ideal circumstances, the dynamic range of human hearing is 140 dB – effectively the difference between a murmur in a soundproofed room to operating noisy machinery or sitting close to the stage at a rock concert. Digital audio is generally said to have around 120 dB of potential dynamic range. In the real world, however, such a wide dynamic range is impractical as music and other produced content are consumed in very noisy environments such as moving cars and busses, crowded bars and restaurants, or busy homes. This is one of the primary reasons broadcasters and streaming content providers process their audio – to reduce the dynamic range and make it more consistent and listenable.
Gain – Technically speaking, a term used to express the ratio between the level of a signal at the input of some circuit or processing stage compared to the level at the output, expressed in decibels (dB). The terms “gain” and “volume” are often incorrectly used interchangeably, but the former usually refers to signal level while the latter refers to a measure of perceived loudness.
Gate – For our purposes in a processor such as Forza, the gate is the control that slows or freezes the action of an AGC when the audio falls below a pre-determined threshold to prevent noise or other unwanted audio from being increased to the same level as the program audio.
High Pass Filter – A filter that allows audio with a frequency higher than its cutoff point to pass through, while preventing audio with a lower frequency from doing so. High pass filters are typically found at the input of an audio processor where they are used to prevent very low frequencies (which have no musical value but whose presence could cause undesirable downstream artifacts and mayhem) from passing through.
ITU-R BS.1770 - The recommended algorithm to measure loudness as presented by the International Telecommunication Union. Their full document is available from the ITU website.
LKFS - The method of measuring loudness in the ITU-R BS.1770 recommendation. It indicates loudness units ("L") referenced to digital full scale ("FS") using the K-weighted measurement curve ("K") which approximates how the human ear perceives loudness. One LKFS loudness unit (commonly abbreviated as "LU") is equal to 1 dB. Though the acronyms are different, LKFS is identical to LUFS, and the terms are used interchangeably in different parts of the world.
LUFS - An abbreviation of "Loudness Units Full Scale," and used interchangeably with LKFS.
Limiter – A type of dynamics processor which typically uses faster attack and release times and higher ratios compared to a compressor or AGC. Limiters with very fast time constants and an Infinate:1 ratio are generally referred to as “peak limiters” as their job is to prevent audio peaks from exceeding a certain level. Limiters can also be used to increase program density (“program limiters”) or work on only certain parts of the audio spectrum (“multiband limiters”).
Limiting – The process of reducing dynamic range by lowering the levels of audio peaks but without raising low-level audio. In the example below, the top waveform is the original unprocessed audio. The bottom waveform has been passed through a peak limiter.
Loudness – In audio applications, loudness is generally used as a comparative term between two songs, stations, or streams. Loudness levels can be quantified by using a meter employing the LUFS (sometimes referred to as LKFS) loudness measurement unit as standardized in the ITU-R BS.1770 measurement recommendation. Loudness also has a qualitative element, as “perceived” loudness plays a role when comparing two signals which may show the same LUFS value on a meter but be experienced very differently. For example, in the files below, the top waveform shows the original, unprocessed audio which had a measured LUFS of -17.28 dB. The bottom waveform shows a heavily compressed and limited version which sounds significantly louder (and subjectively much worse) but it too measures -17.28 dB LUFS.
Low Pass Filter – A filter that allows audio with a frequency lower than its cutoff point to pass through while preventing audio with a higher frequency from doing so. Low pass filters are typically found at the output of an audio processor where they are used to roll off high frequencies, as this is the range where the artifacts of low bitrate codecs are most often audible.
Multiband – A type of AGC or limiter that divides the audio spectrum into several different bands or frequency ranges. Multiband processing allows the spectral balance of the output audio to be very consistent, as each band works independently of the others. This approach also prevents certain processing artifacts common in wideband processing that can occur when a strong signal in a narrow frequency range (such as the “thump” from a kick drum) causes the compressor or AGC to attenuate all frequencies by an equal amount in response.
Presence – The range of the audio spectrum between 2 kHz and 6 kHz. It can add clarity and definition to the sound, but too much of a presence boost can cause an irritating “honky” or “nasal” sound.
Processing Artifacts – By its very nature, processing changes one or more characteristics of the audio, and those changes are rife with compromise. There is no free ride. Much like a balloon, once it is filled with air and tied off, you can squeeze one end to make it smaller, but the opposite end puffs up. Squeeze too much, and the balloon breaks. Below is a list of some of the most common audible processing artifacts and a word or two on what causes them.
Distortion – Technically speaking, any change to the original audio is a form of distortion, but in this case, we are talking about objectionable distortion. Most often we hear distortion as a fuzzy or edgy sound to the audio, not unlike the sound when you turn up the volume too far on a stereo. Sometimes it is subtle enough that only someone who knows what to listen for can hear it. Other times, listeners might not be able to identify it overtly, but be subconsciously annoyed to the point of switching off the music. And sometimes – such as when digital audio exceeds 0 dB full scale, the sound is catastrophically and unmistakably broken. Distortion has many causes, but in general, it is an indication that some component or circuit is being driven beyond its intended range or some threshold has been exceeded. In a streaming audio processor, over-driving the limiters or the bass clipper can sometimes cause audible distortion. Over-driving the input of the streaming encoder can also cause issues, as can a low-quality decoder on the listener’s side of the equation.
Ducking – This describes a sudden rapid attenuation of the existing audio when a new audio component is introduced. For example, imagine an announcer begins a song, lets it run for a few seconds, and then decides to open the mic and talk over the intro. As soon as they begin to speak, the music suddenly disappears to a low level. This is typically caused by fast attack times that instantaneously drive the level of the music down to make room for the voice.
Intermod – More properly called “intermodulation distortion,” this artifact results in some manner of distortion or bad behavior in a part of the audio spectrum outside of the fundamental frequency. For example, if the low band of a multiband limiter is working outside of its ideal range, it may create harmonics that show up in a higher band as distortion.
Pumping – This effect occurs most often in wideband processing when low frequency peaks – such as those from a kick drum – rapidly modulate the rest of the audio spectrum. When a peak occurs, everything else disappears for a split second until it passes, at which point the levels are instantaneously brought back up to their former level. Multiband processing – particularly multiband limiting – can help alleviate this artifact.
Ratio – This is the relationship between a gain change at the input of a compressor or limiter compared to the resulting change at the output. For example, a 1:1 ratio means that a 6 dB change to the input audio level will result in a 6 dB change at the output (which effectively means the processor is being bypassed). A 4:1 ratio means it takes a 4 dB change at the input to change the output by 1 dB (or a 12 dB change at the input to yield a 3 dB change at the output). Higher ratios are described as being “tighter,” while lower ratios are referred to as “looser.” Sonically, a lower ratio will result in a more open and natural sound with a greater sense of dynamic range but may not provide enough control with inconsistent source levels. A higher ratio will provide a more consistent output level-wise but can result in flat sound with no sense of natural dynamics.
Release Time - A setting on a compressor or limiter that determines how quickly it reacts to raise the level of an incoming signal when it falls below a certain threshold.
Spectral Balance – This refers to the relationship of audible frequencies to one another. In the example below, the top graph shows a fairly “flat” curve where all frequencies are at around the same level. The bottom graph shows a curve with a boost in the bass and presence frequencies and a cut in the lower mid-range. Spectral balance is sometimes referred to as the “EQ curve."
Spectral Consistency – The degree to which the spectral balance of the output audio stays the same regardless of the balance of the input audio. Multiband processing is used to actively “re-equalize” the sound so that a bass-shy song gets a lift in the low end while an overly bright recording gets its high end reduced. Some degree of spectral consistency is generally considered a good thing as it helps smooth out variations between songs and provides a more consistent experience for the listener. With some formats, however - such as classical music - it is often considered more important to respect the artistic intent of the original recording, and so less spectral consistency may be preferred.
Threshold – In a traditional compressor or limiter, the threshold is the level the audio needs to exceed before compression takes place. In an AGC, it is the level above which the audio will be attenuated and below which it will be increased. In the case of a gate, it is the level below which the audio will be considered “noise” and not increased toward the target gain setting.
Warmth – A term used to describe a fullness in the upper-bass and lower-mid frequencies (approximately in the range of 100 Hz to 200 Hz). The range of male announcers and singers falls roughly into this range, as does the lower end of an acoustic guitar and the middle range of a bass guitar. A small boost in these frequencies can yield a full and pleasing sound but can result in an unpleasant “boomy” or “tubby” sound if done in excess.
Wideband – A type of compressor or AGC that acts on changes to all parts of the audio spectrum, in contrast to a multiband design that divides the audio into smaller frequency ranges. Wideband AGCs are typically the first dynamic gain stage in a processor and “ride gain” over inconsistent input levels. They also help to consistently drive downstream processing stages, typically the multiband AGCs.
