Issue Number Two
When something like that happens to me, and it has, I usually switch into panic mode. You may recognize the symptoms� sweaty hands, a headache that sits directly behind your eye sockets, and the sudden urge to get out of the video business and join the Peace Corps. Then, after an amount of time proportional to the complexity of the problem, I usually settle down and get to work on a fix. Sometimes nothing can be done to resolve the problem - short of starting over.
At the time, I first went back to the original video footage to see where the problem originated. Upon close examination, I discovered that the 60 Hz hum only showed up in one location; but the person who had delivered the on camera narrative was no longer available to re-shoot the segments. So, I HAD to work with what I had. The hum had been produced by a window air conditioner unit located in an adjacent room. While on set, I was so buried with all the details of the shoot that I neglected to do a location playback of the segments. I had been monitoring using ear-bud speakers which work fine for buzzes and static, but a low frequency hum could not be heard with everything else going on in the room. I also suspect that the built-in compressor on the wireless microphone receiver may have enhanced the room ambience when the speaker wasn't delivering narrative.
Equalizers Demystified
When it comes to buzzes, hums, and other unwanted noises, the single most important tool to the videographer is the equalizer - or EQ as it is commonly called.
Without getting too technical, there are a few terms and concepts you should know about to better understand EQ.
- Frequency is the rate at which a sound wave cycles. The higher the frequency is, the higher the perceived pitch of the sound.
- The amount by which frequencies are boosted (increased) or cut (decreased) is measured in decibels (db). A change of 3 db doubles (+3) or halves (-3) the level.
- Since sounds are usually made up of many frequencies, it usually isn't sensible to change just one. When you boost or cut a particular frequency using EQ, you are also decreasingly adjusting additional frequencies above and below that frequency. The frequency that you adjust is called the center frequency.
- The number of distinct frequencies that can be adjusted is referred to as the number of bands.
- The range of frequencies affected when adjusting a single band is the bandwidth.
There are four main types of EQ - available in both hardware and software variations. All have the ability to adjust the amount of boost or cut of a frequency or range of frequencies.
- Fixed EQ is what you are most likely to find on most home stereo systems - usually with two bands (Bass, Treble) or three (Bass, Mid, Treble). Center frequency and bandwidth are fixed.
- Semi-parametric EQ, also known as sweepable EQ, has fixed bandwidth, but allows adjustment of the center frequency. Mixing consoles often feature sweepable Mid-level EQ in conjunction with two bands of fixed EQ - commonly, Hi (Treble) and Low (Bass).
- Graphic EQ is the same as fixed EQ, but has many more frequency bands. Its name refers to the visual representation of the equalization curve that the multiple sliders suggest (Figure 1). The most common graphic equalizers have 10, 15, or 31 bands. The more bands you can adjust, the more precision you have in controlling the sound.
- Parametric EQ usually provides fewer bands of equalization than a graphic EQ, but additionally allows you to adjust the center frequency and bandwidth of each band. This is the most powerful type of EQ, but is the most difficult for the beginner to master. The best way to use parametric EQ to isolate a particular frequency is to set a narrow bandwidth, with maximum boost, and then slowly sweep the center frequency from minimum to maximum. When you cross over the right frequency, it'll jump right out at you. Then fine-tune the level to boost or cut, and adjust the bandwidth to taste. In software form, you often see the parametric EQ combined with a graphic representation - called a paragraphic EQ (Figure 2).
My experience has been that a graphic equalizer is most often the best choice for videography; while parametric equalization is the better choice for the music studio.
The Software Advantage
The biggest benefit to using software equalization is that you can have much more control over the processing of sound - with less financial investment (assuming you already have a capable computer).
The common mistake people make when using software audio editors is removing unwanted noises by zeroing the sound levels before and after passages of wanted sound (such as dialog). When done without any other processing, this usually results in the noise standing out even more obviously during the wanted passages. There are much better ways.
Keeping to our current dilemma of how to eradicate, or at least minimize, a 60 Hz background hum from a recording, the easiest approach would be to use the software equivalent of a graphic EQ. The (enhanced) screenshots shown here are from my preferred audio editor, Soundprobe. Other popular applications, such as Cool Edit Pro and Sound Forge, have similar functions.
Figure 3 shows a graphic equalizer set to cut frequencies near 60 Hz by 100 db. Because this is a common frequency for processing, this EQ has a center frequency band at exactly 60 Hz. This isn't always the case; so locate the closest frequency in the software you use. With the other bands set to 0 db, frequencies outside the preset bandwidth of the 60 Hz band will be unaffected. This is the simplest way to minimize the offending hum.
Software based signal processors offer capabilities that would be very difficult or impossible to achieve with analog hardware. One such capability is frequency subtraction. This process is related to EQ, but uses an isolated recording of the noise to automatically determine the best processing.
Here's a perfect example. How many times have you shot video in an office building where there was no way to turn off the air conditioning system? Figure 4a shows a voice-over recorded with an air conditioner running in the background. This kind of visual representation of sound is called a 2D frequency graph, and it is a great tool for examining problem areas in sounds. Since the air conditioner is running constantly, we're looking for something in the graph that is also constant. There are three areas that meet this criteria: an obvious, nearly solid, purple line at about 16 kHz; a region of pink in the range of 0 - 500 Hz; and some less obvious spattering of noise over all other frequencies. Sure enough, when we look at an isolated recording of the same air conditioner (Figure 4b) we see all of those areas are present.
Figure 5 shows the parameters of the Frequency Subtractor, using an isolated recording of the air conditioner as the source. Fortunately, it's not necessary to understand the mathematical algorithms used by the Frequency Subtractor to make it do what it does. Visually, it's easy to see that Figure 4a minus Figure 4b, equals Figure 4c. That's good enough for me!
For processing sound like this, Soundprobe has two features that make life much easier. Its Preview function allows you to hear the affect of adjusting various parameters in realtime; and when you're previewing, you can listen to either the effected sound, or just what the effect is doing. For example, in the case of reducing the air conditioner noise, you can switch to Difference preview mode, and hear what is being taken out of the sound. This is a great feature if you are concerned about removing too much of the important sound.
Doing It the Hard(ware) Way
There are many benefits to software signal processing; but if I need to resolve a noise problem throughout a long video project, I would probably choose to use a hardware equalizer. The process is pretty much the same as with software.
There are two probabilities� you either discovered the noise before digitizing the project into your editing system, or you've already edited it and need to deal with the problem after the fact. Ideally, it's the fist case, which means that you can place an EQ between the video source's audio outputs and the editor's inputs, cleaning up the noise before you do the edit.
If the edit is done and you intend to master to tape, it's easy enough to put the EQ between the editor's audio outputs and the mastering deck's inputs. If you're staying with digital for your final edit (outputting DVD, streaming video, QuickTime movie, etc) software processing is going to be the best solution to the problem - allowing you to cleanup your audio clips without having to worry about re-syncing the processed clips.
Hardware equalizers are among the least expensive of audio processing gear, so with a modest investment you can pick up a decent graphic equalizer. One nice thing about EQs is that the technology is established enough that very little changes from one year to the next. Graphic EQs typically come in mono or stereo configurations of 10, 15 or 31-bands, with standard 19" single or double rack space enclosures. Much smaller pedal versions are also available - mostly used by guitar players.
Fiddler In the House
Equalization can be used for a wide variety of noise removal. Other devices such as hiss suppressors, or noise gates, can also be useful tools when striving for cleaner sound. Once upon a time, audio played second fiddle in video production; but this is no longer the case. With 5.1 audio and digital formats, such as DVD, becoming common in every home, videographers and filmmakers need to start improving the quality of their sound tracks. As with most things, there's a lot of trial and error to understanding how best to process audio. Don't be afraid to experiment.
The golden rule of EQ - If you ask a professional audio engineer what the most important thing to know about using EQ is, you will almost always here the same answer, "It's better to cut a signal, than to boost it." Boost does a lot more damage to a sound than cutting does. It's better to reduce the level of all other frequencies than to drastically boost a frequency.
An equalizer is not a filter - Considering the terminology people often use when talking about EQ, it's easy to assume that an EQ can filter out sound frequencies completely. An EQ may make frequencies infinitely quieter, to the point of being inaudible, but they don't completely remove those frequencies.
Always do location monitoring - The best way to avoid having to use EQ, is to get a clean recording in the first place. Before you start recording, try to monitor the room ambience without anyone else around. If you hear any kind of troublesome noise, do what you can do conceal it. If that isn't possible, take the time to make a recording of the room ambience - at least one minute. This can be used one of two ways: either as a source recording to perform frequency subtraction; or as background fill, if you have to re-record audio in another location later and want to maintain consistency in the ambience. When you finish recording each segment with the talent, make the effort to listen to the audio in a quiet area using quality headphones - even if, especially if, you are running behind schedule. Never assume that everything went all right.
Never tell the client about a problem until after you're tried everything to fix it - When I was younger and less experienced, I felt compelled to inform my clients of any problems during production. Now I just go ahead and try to fix it; and if I can't make it better, THEN and ONLY THEN do I tell the client.
(Bob Ketchum, the Sergeant of Sound, has wandered off to make his and her smiley faces on his stereo 31-band graphic equalizer.)
Back to "Online Articles" Page
Back to CCS Home Page