The Music of Losfer Words

Setting proper recording levels and 24 bit audio:
(Copyright 2000 Losfer Words Records All Rights Reserved )

The first step to getting a grasp on digital multitrack recording is understanding dynamic range, decibel levels and audio resolution as they relate to human hearing, which is analog. This is an extreme over simplification of the concepts and does not really explore the deeper math or physics of hearing, but it should help you grasp the concept clearer if you are not familiar with it at all.

The human ear has a dynamic range of approximately 120dB. Every 6dB is equal to approximately one bit of data. So, 120dB divided by 6dB= 20 bits, which means the bit depth quality of human hearing is equal to about 20bit resolution in quality.

Likewise, when recording on digital recording gear every 6dB drop in volume is equal to losing 1bit of a quality. On a 16bit multitrack this means:
at an average level of -6dB, your recording quality is actually 15bit
at -12dB, your recording quality is actually 14bit
at -18dB, your recording quality is actually 13bit
and so on.

The logical conclusion is simply keep your average levels hovering as close to 0dB as possible without clipping. However, this is simply not practical, because it leaves zero room for dynamics (loud and quiet passages) in the music you are recording. The next logical thought would be well record at just above -6dB and you have a 16 bit recording. This isn't correct either, since at -6dB to 0dB the levels are actually varying degrees of 15 bit quality, so the reality is you will never truly create a 16 bit recording without "squashing the dynamics" (compressing the transient peaks (loud passages) of your music down to insignificance in order to keep the levels right at 0dB without clipping). Music without dynamics becomes tiresome and tedious to listen to. Search through your own music collection, find those albums that are loud but that after about a half hour of listening you start to lose interest because the dynamic levels of the music never change.

The same holds true for 24bit recording. You will never actually make a true, full 24bit recording unless your levels are hitting at 0dB without clipping.
However!, you do not want or need a 24bit recording, since the human ear isn't even capable of detecting that resolution.
You want a 20 bit recording.
This means when recording at 24bits you can introduce headroom to preserve the dynamics of your recording and still retain high resolution quality. This will result in music that is more pleasing to the ear and more interesting to listen to. If you bring your average levels down to -18dBfs (-20dBfs is the AES standard for film) you still have approximately just a bit better than a 20 bit recording, (a fraction of a bit more than actual human auditory perception) and plenty of headroom for those loud passages in the music you are recording.

How much headroom?
Ideally you want at least 6dB of distance between your average level (RMS) and those transient peaks.
This means if you average level is -18dBfs your peak levels should be about -12dBfs .

Why no more than 6dB of distance between average and peak levels?
A 6dB increase in level (or 1bit gain) is perceived by the human ear as being twice as loud. So, if your average level was -18dBfs and your peak level is hitting at -6dB, the loud passages in the recording will be perceived as being 4 times as loud. For music that would introduce overbearing dynamics (excessive quiet and loud passages) that would distract the listener from enjoying or experiencing the musical arrangement. It would also be annoying to listen to, as it would induce ear fatigue (excessive vibration in the inner-ear) which most listeners will find either uncomfortable or tiresome to listen for periods of time longer than 10-20 minutes. With your peak levels hitting at -12dBfs you are also leaving headroom for mastering to make your levels hotter without ruining the dynamics of your music.

Why no LESS than 6dB of distance between average and peak levels?
Simple, the dynamics become less noticeable and the music more lifeless. The sonic equivalent of going zero to 60mph and then being stuck at 60mph - boring. If your music composing and songwriter doesn't stink, yet you have been sending out recordings with squashed dynamics or overbearing dynamics you now know why people lose interest after hearing only one song. It is because your message (the music) is getting lost by the messenger (the means of music delivery - your recording). However, we have all made our fair share of such recordings in the digital domain in a misguided quest to have the loudest album. The need to have the loudest recording on the block is actually a carry over from analog recording where you did want levels as hot as possible for cutting vinyl records and FM radio, which would re-compress mastered recordings in an attempt to be the loudest station in town. However, in the digital world preserving dynamics is the name of the game. Especially, with the advent of radio and tv stations converting to digital broadcasting. The need for FM radio stations to be the loudest, and thus overcompress mastered recordings, evaporates with digital broadcasting, and radio will now be all about who has the quality, clarity and dynamics.

Extra info bit depth and sample rates. Use the list below as a guide for setting your target bit depth and sample rate for recording depending on the end goal of your project (Audio CD, Film & Video, or DVD)

16 bit at 44.1Khz - what Audio CDs are encoded at

24bit at 44.1Khz- for multitrack recording for music CDs

16bit or 24bit at 48Khz- used mostly for film and video

24bit at 96Khz- used for converting analog multitrack or master recording into a digital environment, also used on DVDs. If you have the hard disk space available on your multitrack you may wish to record at this depth and rate too. However, be aware that at some point you will have to convert down to 44.1Khz if your music is going to CD audio, which means throwing away about half of those extra digital audio "snapshots" you took at 96KHz.

By recording at 24bit 44.1Khz what you actually will hear as you are tracking and mixing will be a much more accurate representation of what you will wind up with on your final CD which will be 44.1Khz. Human ears cannot detect the difference between 44.1KHz and 96Khz recording, what they can detect is the difference between 20bit and 16bit. And the best way with current technology to get a 20bit recording for analog human ears is to record at 24bit.

What about 32bit float point and 64 bit algorithms?
These are actually interpolations, the recording bit depth equivalent of using an amp simulator instead of an actual tube amp. You want the real thing, not an interpolation or emulation. Tracking at 24bits and the proper levels will give you a true 20bit digital recording, which is just right for analog human ears.