Digital Sound & Music: Concepts, Applications, & Science, Chapter 5, last updated 6/25/2013
Figure 5.23 Finding the smallest possible samples in a digital audio recording
In reality, there is almost always some real-world noise in a sound capturing system.
Every piece of audio hardware makes noise. For example, noise can arise from electrical
interference on long audio cables or from less-than-perfect audio detection in microphones.
Also, the environment in which you’re capturing the sound will have some level of background
noise such as from the air conditioning system. The maximum amplitude of the noise in the
environment or the sound-capturing system constitutes the real noise floor. Sounds below this
level are masked by the noise. This means that either they’re muddied up with the noise, or you
can’t distinguish them at all as part of the desired audio signal. In the presence of significant
environmental or system noise during recording, the available dynamic range of a 16-bit
recording is the difference between 0 dBFS and the noise floor caused by the environment and
system noise. For example, if the noise floor is 70 dBFS, then the dynamic range is 70 dB.
(Remember that when you subtract dBFS from dBFS, you get dB.)
So we've seen that the bit depth of a recorded audio file puts a fixed limit on the available
dynamic range, and that this potential dynamic range can be made smaller by environmental
noise. Another thing to be aware of is that you can waste some of the available dynamic range
by setting your input levels in a way that leaves more headroom than you need. If you have 96
dB of dynamic range available but it turns out that you use only half of it, you’re squeezing your
actual sound levels into a smaller dynamic range than necessary. This results in less accurate
quantization than you could have had, and it puts more sounds below the noise floor than would
have been there if you had used a greater part of your available dynamic range. Also, if you