The digital age of recording requires a far better and uniform system of metering than the purely analog days.
It is now possible to make recordings that have so little background noise that there is very little point in trying to record as loud as possible. Now we have such dynamic range available to us that it would be possible to record the sound of a pistol shot with perfect clarity, play it back at its original level (at 130dBSPL, beyond the threshold of pain) and still be unable to hear the background noise from the recording before the shot was fired.
On the other hand, recording as loudly as possible has become very competetive in the pop world, where material is often played over poor quality media such as night club P.A. systems, In-car systems, FM radio, Myspace, MP3 players and portable equipment. Despite the fact that better quality of sound is available to us, some parts of the recording industry lean toward supplying us with low quality audio equipment and low quality recordings to play.
There is a fundamental question which you may ask, and may have already
asked yourself in relation to recording. That question is how loud do I
mix?
The sound pressure level in different environments will vary greatly
for a given amount of amplification, so how is it possible to have some
reference level for recording, mastering and mixing when we already know that a listner may very well change the level of the sound on reproduction?
In this section I would like to bring to your attention the importance
of the amount of sound in your control room being relative to the level
displayed by your meters.
The actual sound pressure level (SPL) in a listening environment is
something that each listener will adjust to taste. This can and should
vary, but you may have noticed that the loudness of individual
recordings can vary to a great degree. So how do we apply a standard
for loudness which will satisfy musicians, recordists, mastering
engineers, radio listeners, broadcasters, theatre-goers and domestic
listeners?
Many people hook up a cd player and set it to read 0dB on
their master meters, adjust the volume of their control room monitors
to be comfortable at that level and proceed with recording and mixing
down their material. In theory this would appear to be practical, but
there are many further considerations. It would be quite easy to set
your monitors to a louder level because you like the tracks you play
and enjoy hearing them perhaps a little louder than is ideal. It is
also more than likely that the recording levels on the tracks will
vary, making the decision very difficult.
Even before you start recording you will benefit from acknowledging
that it is important to monitor at a certain loudness in the control
room. Above a certain loudness our perception of dynamic
content in audio material tends to diminish. We become less aware of
compression because the Human brain starts to compress the sound we
hear. This phenomenon varies depending upon not only the loudness of
the sound, but also the amount of time the listener is in the
environment. Anyone leaving a loud rock concert or night-club will have
noticed that it takes some time for the brain to re-adjust to a
'normal' level. To a lesser degree this effect can easily happen to an
engineer in a control room. In fact it is no less important to give
your ears a rest when mixing down than it is to take your eyes away
from a computer screen. Clearly, to encourage any sort of insensitivity
to sound
is of no advantage to a sound engineer and to be avoided. Adversely,
attempting to adjust sound whilst listening at a lower than ideal level
can tempt us to make recordings which are too "hot" and lack dynamic
range because we attempt to bring the sound level up to an SPL which
our brains consider ultimately comfortable. This can result in a flat,
un-interesting mix with little dynamic range. Clearly, somewhere
between these two extremes lies a relative listening level and
recording level which should allow us to achieve the best results.
Recordings of popular music have been getting "louder" for many years.
That is to say, the average level of a recording has become closer to
the maximum available recording level. So the perceived loudness is
greater, but the depth of the dynamic, the "liveliness" has decreased.
A higher average level means less available headroom and so a flatter
mix, but compare the actual comfortable listening level of an older
recording and you are likely to find a far more dynamic sounding mix,
but at a lower average level. In other words, the consumer is likely to
adjust the "volume" upwards to achieve the same comfortable listening
level and to perceive a greater dynamic.
Broadcast engineers have to balance the transmitted output from TV and
radio stations to provide a regular relative transmission level. Moving
picture theatres need to provide a sound level which has a far greater
dynamic range than a Pop CD. The sound is likely to have a far lower
average level and greater headroom available for special effects etc.
In other words, it is necessary to consider the circumstances under
which the media will be reproduced as well as the content of the media
itself.
Certain commercial entities have taken advantage of the fact that no
standard exists. It has become widespread practice to make recordings
with a very "hot" average level to make advertisements which sound
louder than an average film track. In the Pop world a "loudness war"
has been going on for some time. Despite the fact that as digital
recording technology has improved, higher quality recording has become
possible, certain parts of the industry have actively attempted to
reduce the quality of output in favour of greater average level and as
greater bit depth and sample rate increase the quality of digital
recording and yet greater dynamic range becomes possible, the industry
may seek to continue the loudness war and make the problem of relative
sound levels even worse.
So how can a recording engineer decide on a relative listening level
which will result in a recording acceptable to the industry and the
domestic listener, given that the end-user will adjust the level of
their system to taste? how can we produce appropriate content for
delivery via different media?
Currently there is not an absolute standard, but much work has been
done in considering the problem and certain standards are widely
accepted thanks to a proposal made by the experienced mastering
engineer Bob Katz.
It is clear that to provide an outline for an "average comfortable
listening level" much experimenting would need to be done. In practice,
engineers in the film industry have already carried out detailed
subjective listening tests in this area and regularly use a mixing
environment containing a fixed, calibrated monitoring system set for an
SPL of 83dB. This is considered too loud as a domestic listening level
because most home listening situations would not accommodate such a
great dynamic range, so many authorities recommend lowering the gain by
6dB. Bob Katz has experimented over the production of many masters to
develop a consistent perceived loudness when creating CDs. He has found
this level to be consistent with the recommendations and has proposed
not only a standard for an average monitoring level, but also "K
Metering" a system of metering which refers to program level depending
upon the intended use of the medium.
The wonderful thing about the K system is that it doesn't stop
at metering.
Consider for a moment the different sorts of dynamic information which could characterize a recording. They fall broadly into three groups.
Bob Katz, a leading mastering engineer proposed that there should be not one but three standard levels for recording in the digital domain, each of which could best reflect the needs for effective delivery of a particular media group.
In recording movie sound-tracks for example, the spoken dialog needs to be intelligable, but reproduced at a level whereby it is in harmony with the backing track and the sound effects. Preseving the reality of the sound effects will often require that they are recorded with a great deal of headroom available so that the dynamic can be preserved.
A classical or operatic concert is a large scale acoustic concert with a dynamic which will vary between the sound of a single flute and the whole orchestra in unison.
In the case of Pop and Rock music compression often constitutes some of the character of the music. The sounds are very complex and are often changed so much that any solo'ed component would not sound as though it belonged to the whole project. The average level of the music is lightly to be higher, but the requirement for headroom is lightly to be much lower.
In the case of media for broadcast, the stations themselves compress the transmissions to obtain a constant, clear signal. In these circumstances, by decreasing the dynamic range of the original recording it may be possible to ensure that the master is better represented across the transmission.
The learned Mr Katz cites research which indicates that over a large representative audience in a theatre (over 1000), a certain reference SPL is considered not to loud and not to soft for the audience. This magic number is 83dB SPL. He then goes further to cite references from respected audio engineers who recommend reducing this level by 6dB for domestic listening.
With reference to 0dBfs the 'K' system recommends are average recording levels of:
The image below demonstrates how the meters appear compared to a normal digital meter
Bob Katz' full explanation of the K system is available here.
His original proposal and the reasoning behind it is here.
The K metering system is of great benefit to an engineer. It
provides:
A consistent monitoring level
The opportunity to create mixes which will contain the desired dynamic
content when reproduced and distributed using specific digital media.
A consistent recording level for a "comfortable" listening level
A consistant metering system to comply with the required loudness for
CD, DVD, Film and Broadcast production
It is not complicated to calibrate a monitoring system to comply with
the K system.
To set up the SPL of a monitoring system:
Play a standard pink noise calibration signal at a level of -20dB FS
RMS according to the digital peak meters through one monitor at a time.
Adjust each one to a sound pressure level of 83dB using a calibrated
sound pressure meter set to C-weighted, slow response.
This -20dB point on the peak meter now represents an average
level of 83dB, the standard for the film industry.
Now you can set the "0" level at 6dB lower on the meter to represent
the suggested level for domestic monitoring at 77dB.
There is a free VST plug in named MDA test tone which provides pink and
white noise, test tones and a sweep generator. It is available at
Smartelectronix.com
The 'K' system is designed primarily for allowing us to create our final mix-downs at a a suitable level for their intended use, but what about the level at which we record each track in a multi-track mix?
There is much discussion about the marvels of recording with software that uses 32bit-float and 64bit-float audio engines. It is often presumed that this enables us to record at any level and that it is impossible to obtain distortion, so one might as well record as loudly as possible. This is not the case.
Digital recording software of this nature will unlikely distort your recordings. It is almost impossible, but as with a "real" studio, it is important to ensure that the amount of gain, or loss in the audio chain remains relatively constant through the various stages of your DAW software. Many FX plugins may not have the same large bit-depth as the application which is hosting them. This being the case it is quite easy to cause a plugin to distort the signal, especially at the point between one effect and the next in a chain.
Since most recording software will provide unity gain between a channel and the main output with no adjustment this also allows the user to make good value judgements about the qualities of each sound with the same accuracy which would be applied to a final mix-down. This can also be applied to sub-mixes.
It is quite common to render tracks and submixes to an audio file in order to free up processor power for more complicated mixdowns and allow the use of additional plugins. These audio files may be re-used as stand-alone mixes for some other purpose or inserted into tracks with different FX applied, so recording them at a sensible level is just as important. Remeber that the noise floor for each audio file will be very low. This is of course subject to your recording skills in the use of the analog equipment connected to the input of the sound card. Many a bad, muddy sounding recording is blamed on one or another DAW system when the user has not paid enough attention to this critical point in the recording chain. Digital recording is very powerful, but it's not magic.
In real terms, I would suggest that the best approach is to treat each and every track as if it were a master track, using the lessons of the 'K' system to your advantage. Pay as much attention to clarity and dynamic range as you can. This way you can quickly build a sample library of outstanding quality, the flexibility available to you in the mixing process will be greater. You will have greater control over your mixes and they will sound sharper and more apparent. You will far more easily be able to create the required amount of space in your final mixes if you place the audio comfortably in the space available, considering both the space between the noise floor and the quietest detail of each sound and leaving plenty of room for the highest peaks.
This site is under on-going development. More pages will be added soon. Thanks for visiting inthepipeline.net