In April of 1997, I wrote an article for NewTekniques Magazine called "Getting better Flyer sound". In the article I briefly touched on several ideas which would help improve the audio capturing capabilities of the Flyer, including a paragraph on using compressor/limiters to get a more even audio track.
An outboard limiter will help keep Flyer audio clips from being corrupted due to overloading the audio card during the digitization process. It will establish a maximum audio level while maintaining a minimum distortion level . By now, we all know that perhaps the Flyer's greatest inherent weakness is the inability to allow the user to adjust gain parameters while digitizing an audio clip. In other words, once you've started recording, you cannot change the volume one way or the other. A compressor/limiter, when properly adjusted, will automatically "turn up" the volume on quiet passages and "turn down" the louder ones.
Ozware's "ShortCuts" software has directly addressed this problem by allowing incremental volume changes to the audio during the digitization process. While this is a definite improvement to the original Flyer design, it is not exactly the pot of gold at the end of the rainbow. The operator must still make decisions all throughout the process if there are wild differences in audio, which is the case more often than not. That's why I still recommend purchasing a dual-channel limiter to insert in the signal chain just before the Flyer audio inputs.
A recurring thread on the VTFML is just HOW to properly set and use the limiter. Another area of confusion stems from the differences between a compressor and a limiter. Many compressors give the impression that they will also act as a limiter, but that is not exactly true in our application. Remember, as a rule most limiters and compressors are aimed squarely at the music market. They are designed to be used in either recording music or in sound reinforcement. Sometimes all those "extra knobs" found on many high end devices are overkill for our purposes.
While compression is a useful tool in audio recording, the limiter is what we really need to keep our Flyer audio clips safe and sound and glitch-free. But what is the difference and how do you set these things up for digitizing audio to the Flyer? These are a few of the topics we will cover in this article. Let's start will an overview of the feature set found on most compression/limiting devices.
By definition, a Compressor/Limiter is an electrical device, which prevents a linear increase in output level beyond a pre-determined level (the threshold). This device does not exhibit linearity between input and output levels due to a circuit which reduces output by a ratio of the input signal exceeding a set amplitude. In more simple terms, it is an electrical device that controls or affects the dynamics of an electrical signal. The effects resulting from dynamic processing are termed COMPRESSION and LIMITING. A Compressor is a variable gain amplifier whose gain decreases as its input level increases past the threshold point. A compressor may use low compression ratios of approximately 2:1 or 4:1. A Limiter is a variable gain amplifier with a high compression ratio of approximately 8:1 or greater. The high ratios maintain essentially constant output level despite increases in input level above the threshold.
A Compressor/Limiter can be used to control a signal so that is heard without becoming too loud, thereby distorting or damaging the system or hurting or offending the listeners ears (n the case of a public address system) . It can also be used to control a signal for recording and broadcast so that it does not become so soft that low level electrical noise such as hums, hiss, etc. are at a similar level. Furthermore, you can use it to control a signal by reducing its dynamic range and increasing its average level relative to other signals to give it more prominence over the other signals. It stands to reason that a device so flexible to be used in so many applications might be more than a little confusing to anyone trying to figure out how to best set it up for a particular application, like digitizing audio footage into a computer. And labels like "Attack", "Threshold", "Release", and "Ratio" might as well be Greek. Someone seeing the term "Soft Knee" might wonder what this all has to do with the medical profession. So, next �.. let's discover what these names really mean.
COMPRESSION is the general term given to something that dynamically reduces the dynamic range of a signal. It occurs when the average input signal level is above the threshold setting. Too much compression applied to a signal will result in extreme amounts of background noise, or the quieter passages being "sucked up" to a maximum volume setting. Thus, NewTek's advertising department has (presumably) unwittingly named the beast in their recent magazine ads which show an unfortunate Opossum with highway paint sprayed right on top of the road kill. The caption for the ad reads: "Compression Sucks!" How true! Unfortunately, that describes two different processes. "Audio compression" NOW describes the act of compressing the audio signal AND compressing computer data (as in Internet usage). But that's another story����.
LIMITING occurs when the average input signal level is below but peak signals exceeding the threshold are attenuated. This may also be termed "Peak Limiting". The effects on the signal are for only as long as the circuit is acting upon the signal. This is determined by the time and degree to which the signal is above threshold and the way that the circuit reacts to this. The device's circuit will be optimized to act upon either average or peak voltages. There may be a combination of both circuits in one unit or configurable to either.
In simple terms this means that a limiter may be configured to allow all the dynamics through the signal chain up to a level set by the user. Any signal over that set level will be "limited" to avoid exceeding that level. THIS is the main function that we are interested in for digitizing to the Flyer. Those undesirable peaks will overload the Flyer inputs, resulting in the dreaded corrupt clip. And as we know corrupt clips WILL cause sequencing errors in a Flyer project.
Continuing in our "Compression Primer", the following are a number of definitions, descriptions and terms relating to the operations and descriptions of knobs and buttons found on many COMPRESSORS and LIMITERS:
Compression Ratio - The ratio, in dB (decibels), of input level change to output level change, above threshold. A compressor whose output level changes 1 dB for a 2 dB input level change, above threshold, has a 2:1 Compression Ratio. A compressor whose output level changes 1 dB for a 4 dB input level change above threshold has a 4:1 input level change above threshold has a 4:1 compression ratio, and so forth.
Threshold - The level at which onset of compression occurs. When the input level is below the threshold setting the signal is unaffected. When the input level exceeds the threshold there is attenuation of the output signal relative to the compression ratio and the degree to which the threshold is exceeded by the input.
Time Constants - The speed that the device reacts to the dynamics of the input signal (envelope) is set by its attack and release times.
Slope - The ratio of output to input gain.
Envelope - The graph of a signal from start to finish. Attack, Decay, Sustain and Release.
Level - Refers to both sound waves and electrical voltage waves. Is the mathematical average of the positive peaks.
RMS (Root Mean Square) - The mathematical process to derive the averaged value of level.
Peak - Highest level of signal. Signals with very fast attack time, high amplitude and fast decay are also called transients.
Dynamics - The movement between the lowest and the highest point. In sound, the softest to the loudest.
Electrically, the lowest to the highest voltage.
Dynamic Range - The difference between the lowest and the highest point. May be referred to peak levels or to RMS levels. The peak level dynamic range will be greater than the RMS.
VCA (Voltage Controlled Amplifier) - An amplifier which reduces its gain according to the amount of input received. Control is by an external voltage. In application as a compressor/limiter the input signal is sampled and that sample, when of sufficient magnitude (i.e. exceeding the threshold), will cause the output level to be reduced. This processing is done via a side chain.
Side Chain - A control circuit outside of the signal path. A sample of the signal is taken and dc voltages are derived to control a VCA which acts upon the signal output. In other words, you use the side chain in triggering the processor from the voltage output of an external source. For example, you could drive the device (via the sidechain) from an equalizer, resulting is turning the sidechained compressor
into a "de-esser" by selecting the offending frequency on the EQ. This, in turn will result in the selected frequency being de emphasized in the final audio track. Or you could tighten a bass guitar track by sidechaining from the kick drum. Each time the drum is hit and the signal exceeds the selected ratio, the bass guitar track is opened and shut, dictated by the length of the drum's volume envelope. You could also use an equalizer to de emphasize the background noise in a narration track by using side-chain operation.
Detection - The method of measuring signals exceeding the threshold level. Usually Peak detection is used for Limiting and RMS for Compression, but both work as either depending upon the nature of the input
signal and its level in relation to the threshold.
Concerning the application of Compressor/Limiters:
For a Compressor/Limiter to operate there is required a signal level greater than the threshold setting. What occurs after that level is reached is subject to the character of the input signal, the degree to which it exceeds the threshold and the operational parameters of the VCA circuit.
The input signal may be simple or complex, that is, be a mix of many varied signals or just one single
sound source or `voice'. A sound will have an envelope that typically will commence from silence then rise to its peak then return to silence. In the real world pure silence is rare. A universal rule in audio engineering is that a signal at any part of the circuit should be at as high a level as is possible without overdriving the electrical components with high voltage. This provides the greatest signal to noise ratio.
The maximum level is determined by the manufacturer and is expressed in decibels of headroom above a specified nominal voltage. A Compressor/Limiter can be configured to prevent any voltages reaching this circuit maximum level. By increasing the gain of the signal at this point the average signal level will be higher and therefore the signal to noise ratio improved.
The degree to which the signal exceeds the threshold of any type of Compressor/Limiter circuit and the effect there obtained will be determined by the character of both the signal and the circuit. The action of the signal input increasing strength and exceeding the threshold causes the VCA to begin to attenuate the signal output.
The character of the transition of the signal past the threshold could be classified between extremes of
being percussive or non percussive. A percussive sound has a fast attack, a non percussive sound has a slower attack. It is the rate or slope of the signal attack and the proportion of signal above threshold that will affect the action of the VCA. The more percussive a signal is, the faster is it's ATTACK rate. If it's decay is also fast, a fast attack from the circuit would be required to affect any attenuation on the signal.
Many devices have a gate built into the circuitry. The gate can be set to allow any sound source, when reaching enough amplitude, to trigger (open) the gate so the signal passes through it. Anything set below the desired mark will not allow any signal to pass through the device, in effect closing the door to the audio signal. This application is great for cutting off unwanted "walla" noise (ambience, etc.) when a speaker is not talking. Like many operations of this complex device, a little processing can go a long way. For instance, if the background noise is very loud, using a gate can actually work against your purposes, resulting in complete and uncomfortable silence between passages. The aural impression to the listener is that there is a "short" in the audio source. A gate is only useful for intentionally and completely cutting off the audio signal.. For instance, you could gate all the drum microphone so they only let sound pass ONLY when the drum is actually hit. It's perfect for taking out drum overtones and sounds eminating from other sources. If, for instance, you videotaped a politician making a speech at a zoo in front of the monkey cage (appropriate, yes?) and the monkeys were screeching horribly through the presentation, you could use a gate to cut off all the audio when the speaker isn't talking, but it sounds so UNNATURAL when you have DEAD SILENCE it's as obvious and distracting as the screeching simians in the first place. For Toaster/Flyer audio purposes, the gate is not a necessary item unless the background noise is very low and you might possibly have a music bed running underneath the narration.
Many Toaster/Flyer operators have asked me what the "proper setting" is for digitizing audio through a limiter. There are many variables. The type of device you have makes a lot of difference. Not all limiters and compressors are created equal. Generally speaking you should look for a device which advertises full limiting and has at least a PEAK STOP button. I will refrain here from name dropping as there are many good processors in the market by several manufacturers. Do some shopping. Make comparisons. Don't just drop down the dollars because it has a lot of knobs on it.
Then there's your particular application. Some people just concentrate on weddings, where there is a lot of application for a limiter. Weddings are tough to document anyway, and proper audio can be of paramount importance to your client. You can't exactly stage a wedding over because you "lost" something, so getting it right the first time is critical. With a limiter and/or compressor, you can at least "fix it in the mix" to a certain degree. Any kind of "real world" event recordings (speeches, events, celebrations, etc.) could benefit from some dynamics processing inserted in the audio signal chain when digitizing to the Flyer.
It takes some tweaking but when you learn what a compressor/limiter does you can quickly set it up on a "per job" basis. The PeakLimit knob keeps the Flyer from getting clipped at the top end, and the compressor will "suck up" those quiet moments and bring them up to 0VU for you. In most cases I do not desire a compressor (camcorders usually have onboard compression of some type anyway) but I use the PeakLimit a lot! It keeps the signal from hitting that red line on the Flyer.
Remember the Ketchum Credo: Redline = sequencing glitches.
And now a brief explanation of the audio files and screen grabs accompanying this article:
As already stated in this article, the limiter/compressor is such a flexible processor, and it's applications are so varied, that to simply state "turn this knob here and press this button" does no good as a explanation of how this device works in any given situation. By tweaking even a single parameter, the operator can make such a subtle change that it is virtually unperceptable to the human ear. Likewise, a single knob twist can produce an enormous change in the original audio signal. The real trick is to learn by doing which parameter or combination of parameters affects the signal in producing the desired effect.
The audio files and screen grabs have no real purpose other than to show by example what the results of "knob tweaking" can produce. Also, since I used software based processing in the preparation of this article as opposed to hardware, "knob settings" are interpreted in increasing/decreasing numerical values. My software of choice for these types of audio processing is Soundprobe, which is now in the Revision 8 of Version 2. Most audio software is similar in form and function, but Soundprobe offers the added benefit of allowing you to actually listen in real time how your tweaking affects the signal without first rendering the effect to the file. This saves an enormous amount of time.
The following descriptions will give you a better idea of what your looking at and listening to. What you will have to determine is how my software settings correlate to your hardware device.
ORIGINAL FILE (Kit_Clean): This one is pretty self explanatory. I chose to use a segment of a drum kit loop. Drums are perfect to use for examples because they have extremely loud and soft passages and cover a wide frequency spectrum. The cymbals are a bit distorted by design for the purposes of this demonstration.Read on, and use this screen grab as your watermark when comparing to the processed files.
COMPRESSED (Kit_Compressed): You can tell when listening to the audio clip that this clip is pretty squashed. The kick drum has very little presence and sounds about as loud as the snare hit. The cymbals are splattered and actually louder than the rest of the kit. I achieved this effect by setting the attack at 400 ms, low onset at -29 db and high onset at -2 db, added 8 db of expansion and -12 db compression, 8 ms of recovery, and set the threshold to 2db. I would not call this result pleasing, but by example it could help a narration track that suffers from lack of dynamics.
SLIGHT EXPANSION (Kit_Slight_Expansion): With this setting, which is more subtle in approach, you will notice that the kick and snare drums still have some dynamics, and the cymbals are less splattered. Input level is set at 1.5, -29 db low onset level and -55 db hi onset level, 400 ms attack, ) db compression, ) db expansion, and 1 ms recovery time. You can see by the screen grab that the waveform is a bit wider but the final peaks are about the same as the original file.
OVERSATURATED (Kit_Oversaturated): This setting is waaaaaaaay over the top! It has no practical purpose other than to show by example how you can go too far in knob tweaking. The level is set to 4, low onset at -1, high onset at 0 db, attack at 1060 ms, compression at 0 db, expansion at 12 db, and recovery set at 1 ms. A glance at the waveform will verify that the signal is going to sound like a perpetual explosion, and the sound clip will testify to that!
LIMITED (Kit_Limited): This is subtle but very effective. First, you will notice that this setting has actually reduced the cymbal splatter considerably, making the audio clip sound more uniform and clean. Yes, the overall volume is a bit lower, but the peaks are manageable and contained. All you need to remember is to set a slightly higher input level to the Flyer. This is the main reason why I advise limiting over compression when digitizing audio into the Flyer. To achieve this effect I have the level set at -10, low onset at -100 db, high onset at -100 db, attack at 100 ms, compression at -20, expansion at 1 db, and recovery at 1 ms.
Bob Ketchum, the Clowned Prince of Audio, lives by the credo:
"Audio compression GOOD�. Video compression BAD� Spinal compression TERRIBLE"
Back to CCS Home Page