8 Audio in Post Production
When dealing with audio in post production, the aim is to achieve a well-balanced and audible soundtrack mix of three elements. Firstly, the ‘Feature’ sound (the main audio that you want your audience to hear, such as the dialogue or voice-over); secondly, the ‘Ambient’ or background sound (the audio which will help establish the scene and environment); and, thirdly, the additional ‘Effect’ sound (such as a music track or additional sound effects that help to tell the story or develop mood). Balancing or mixing all or some of these three elements will enable you to immerse your audience more fully into the video and enhance the images.
The aim in the post-production stage will be to non-destructively add these three elements together to make your completed soundtrack mix. However, just as there are ways to improve poorly recorded sound, it’s also possible to reduce the quality of correctly recorded sound in post-production. Simply adding the feature sound and effect sound together without thought to comparative levels and audio frequency control will more than likely result in a confusing or indistinguishable soundtrack. So it’s not just a matter of laying the whole lot down without any thought. There are a number of processes and skills to learn.
8.1 Monitoring Sound in Post Production
The quality of your audio monitoring during post production will have an impact on the final soundtrack quality. Using PC-type speakers attached to your editing suite will, most likely, not give you a true representation of the tonal qualities of the audio. This is because, like headphones, speakers are designed for specific purposes with particular characteristics.
Cost plays a big part too, and you tend to get what you pay for. Cheap PC speakers will generally have little in the way of low frequency reproduction (bass) and will accentuate the top end frequencies (treble). Hi-Fi speakers will reproduce sound with more warmth and clarity by boosting the low and high frequencies.
Adjust the tonal qualities of your soundtrack so it sounds correct on either of these speaker types and it might not sound the same when played back on your clients TV (although that is true no matter what speakers you use for monitoring, because you can’t always know how good, bad or indifferent your client’s sound system will be.) However, it helps to start by listening to speakers that are as “neutral” as possible.
Professional studio speakers (known as ‘Monitors’), whilst more expensive than Hi-Fi speakers, will produce audio signals that are a more accurate representation of that which was recorded, reflecting the dynamics and frequency responses of the microphones or pre-recorded material. They are also designed to deal with overloading; peaks and feedback which will eventually damage or affect the performance of both PC and Hi-Fi speakers.
Whilst studio monitors provide you with un-enhanced, robust and more accurate audio reproduction, it is unlikely that your clients will be using these when watching your programme. It is important, therefore, that you check how your soundtrack will be reproduced on other sound systems, including Hi-Fi speakers and speakers within TVs. Use a selection of these different types of speakers to test how your finished production will sound.
As we mentioned back in section 1.1, human hearing has a different “frequency response” at different power levels, so at the final checking stage try listening at a range of volume levels and make sure that your Feature sound is always audible. However, whilst mixing your audio tracks and setting levels, try to always have your volume control (on the speakers or amplifier) at exactly the same setting (mark it with a pen).
The idea is to establish a personal “reference level” that you can get so accustomed to that you will be able to tell more or less whether your signal levels are correct without looking at the meters, and when jumping from one section to another will be able to tell whether they match in terms of level.
8.2 Audio Levels and External Metering
Adjusting audio levels is an important stage in audio post production; not only the overall audio level of the master soundtrack, but also comparative levels between the three main elements of sound (Feature, Ambient and Effect). Setting the overall level too low will result in the viewer having to increase the volume on playback, which will also increase unwanted electronic noise generated by their playback system. Setting the overall level too high will result in distortion, or worse still, may damage the viewer’s playback system (see section 6.3 – Correct Recording Levels).
Mixing sound elements together effectively requires particular attention to their comparative levels. If you are adding a voiceover track and the level is too low in comparison to the ambient or music track, then it will be indistinguishable to the viewer. To compensate for this the viewer is likely to increase the volume on their playback system and that will result in the ambient sound or music becoming distorted.
It is usually best to start with the main Feature sound tracks, and adjust the gain so that peak dialogue levels are in the range – 18dBfs to –10dBfs. Short peaks above this level can be controlled with a limiter (plug-in), but any levels that fall below this range for more than a few seconds will probably need increasing.
Once you’ve established consistent levels for your Feature sound, you can start adding the M&E (Music & Effects). As general rule, the maximum level for music should be set slightly lower (at around –14dBfs) on the meters than speech, because it tends to contain more energy than speech when both have the same peak level, so it sounds louder. Of course if your music is playing at the same time as a dialogue track, the level will have to be reduced considerably so as not to overpower and distract from the dialogue. Note that it’s the music level that should be decreased, not the dialogue level increased (which you’ve already set before fading up the music), in this sort of conflict.
Similarly, add in your sound effects and ambience tracks, keeping them at a level where they don’t overpower the feature sound, but are audible enough to do their job. The generous use of audio cross-fades, fade-ins and fade-outs, even where the vision uses straight cuts, will help the illusion of overall continuity by blending the sound track into a continuous entity. The M&E tracks are very like seasoning in cooking. The judicious addition of a little salt and pepper, herbs and spices can enhance the overall flavour, but too much of any of them can completely ruin the dish. That said, rules are there to be broken, but only if you understand what they are and why they exist, and know why you’re breaking them.
Finally, it’s worth mentioning the importance of good audio meters. Good speakers will help your ears assess the quality of your audio, but until you’ve managed to calibrate your brain you’re best not depending on it to objectively help you set absolute signal levels. Most audio and video editing software has some sort of audio meter display, but many of them do not match he requirements of an audio meter for video production. Having a meter which gives readings down to –60dBfs or below is of no practical use to anyone outside a lab measuring signal to noise ratios. The area of interest is the top 30dB, in the middle of which you’re trying place your Feature sound. Anything below –30dBfs is likely to be background sound, and you don’t need a meter to tell you how loud it is (because there’s no risk of distortion at that low level), just your ears to tell you how it fits in subjectively with the rest of the soundtrack.
What you need the meter for is to make sure your (louder) Feature sound is loud enough but not too loud, so you need a meter which takes that relatively small range at the top of a linear scale and magnifies it across most of the meter display, making it much easier to concentrate on. At the same time, you don’t want a meter display that jumps about too much as the levels change, which equally makes it difficult to follow. What you’re interested in is the peak levels, to avoid overloads, so an ideal meter would show the peaks whenever they occurred, but only fall back down the scale far more slowly, to give the user time to appreciate how high the peaks were.
8.3 Graphic Equalizers and Audio Sweetening
The equaliser is a powerful tool, and when used properly it can greatly improve the quality of the sound you record. Simply put, a graphic equaliser is a device used to reject (cut) or boost (raise) a series of predefined frequency ranges or bands. Equaliser come as either standalone hardware items, as part of an audio mixer, or as software programs within nonlinear edit suites, or as separate audio treatment programs. Equalisers are used to target specific frequency issues, which is why the more bands you have, the greater control they will provide.
It is important to remember that equalisation is the last step taken when we are tuning our soundtrack. The purpose is to help produce a natural sound that is pleasing to the ear, and and ‘sweeten’the sound we record – but if used incorrectly an equaliser can spoil your soundtrack completely.
The best way to appreciate the effect of using a graphic equaliser is to experiment in adjusting the various frequencies and listen to the effect this has on the overall sound quality. Using a familiar music track, try raising the level of each band one at a time. As you do so you will notice that a change takes place in what you are listening to. The sound you produce will, if you’re working from left to right on the equaliser scales, go through different stages. It will begin as a rumble; then become boomy; then boomy and boxy; then boxy; then nasal; then metallic; then have ‘ssssss’ sibilance; then sizzle; and then become undetectable.
Now apply this process to your own recording. When you find a band that when raised becomes unpleasant to listen to, mark that as a band to cut or reduce. The more unpleasant it is, the more you cut it. As you experiment with these bands you will find that it leaves a pattern in the band sliders hence the ‘graphic’ in the title.
If you are trying to improve voice reproduction, such as a voiceover or person speaking to camera, a slight raise to the 2kHz or 2.5 kHz band can often help. Equally, a slight raise around 8kHz-9kHz can improve clarity. If there is too much around the 200Hz-300Hz range the result can produce a muddy and indistinct voice, so a cutting of these frequencies will help. Avoid the temptation to simply set the standard ‘smiley face’ on the equalizer, as this takes down all of the mid range frequencies where the main signal of the human voice is, leading to just a weak sound.
Remember that raising a frequency or band is adding noise to your recording so it should be used as sparingly as possible. If you believe you have a problem in your soundtrack, but are not sure where it is, try boosting the frequencies slightly one at a time. When you hit the problem area the effect will be magnified and should become noticeable. Apply a slight cut to that band and the problem should be solved. If you find that you are cutting many frequencies the issue may be that the signal is just too loud. In which case you should use the volume control and not the equaliser.
Finally, skillful equalising is not about rules – it is about listening analytically to a sound and adjusting to suit.
8.4 Dynamics processing
Along with audio EQ (equalizers and tone controls), dynamics processors are probably the other most useful tool at the audio post production stage. Of course AGC is a type of dynamics processing, and the suggestion was to avoid using it, so what