- spencerm96

# Sample and Bit Rate: WHAT ARE THOSE!?

Updated: May 28

Hi, it's Spencer Miles the owner of Spencer Studios, which is a recording

studio in downtown Lancaster Pennsylvania. Today we are answering some

questions that nearly everyone has wondered at some point in their musical

endeavors. When you enter any current modern commercial grade DAW you are

presented with two questions if not among others. You are asked what sample

rate and what bit rate you would like to use. Today's article will help you

make an informed decision when presented with these questions. This topic

can go greatly in depth however, we are going to keep it at a consumer level

for the purposes of this blog.

First let's define each of these terms. Sample rate can simply be described

as how often the computer is recording data. The measured amount of

recording known as frequency, is X number of times per second the computer

writes and stores data. An example would be 44.1 khz or 44,100 times per

second. Bit rate can be thought of as the amount of data that is recorded in

each sample. Specifically in audio, the data constraints are pitch

(frequency) and volume. Sample 1 of 44,100 recorded on Sunday at 1:01:05pm

might have recorded a trombone playing a fundamental pitch of 100hz with a

positive amplitude (similar to volume) of +4. It will have have also

recorded the overtones of that fundamental and associated amplitudes of each

overtone frequency at the same time. The construction of the overtone series

is what gives each noise we hear a unique sound. One can think conceptually

that the computer uses these samples on a multidimensional graph of sorts to

reconnect the dots thus recreating a recorded pitch. The rate of recordings

and amount of information recorded in each bit is what controls the accuracy

to which the computer is able to reconstruct the recorded sound. If we were

to create this 3D graph the X would be time, Y amplitude, and Z pitch

labeled as frequency.

Why does the sample rate in most cases start at 44.1khz? In audio we have to

acknowledge the Nyquist theorem which says to accurately recreate a sound we

must sample at twice the frequency in pitch. Human hearing generally exists

from 20hz to 20khz, so we have to sample at least 40khz or 44.1 to give some

wiggle room. Let's look at the below drawing as an example. Imagine we are

recording a sine wave which is a pure tone existing at one specific

frequency with no overtones. The y axis is amplitude and the x axis is time

in samples. Imagine this sine tone is at 20khz and we are sampling at 40khz.

Imagine the sine tone in black is a perfect curve and not hand drawn. The number of samples on the x axis are labeled corresponding to a red dot as the specific sample.

When the computer connects the dots above it is still able to recreate a

sound with the same volume and pitch as seen in gray but with slightly

reduced fidelity. Now let's see what happens if we sample less than twice

the twice the frequency below.

With too slow of a sample rate, when the computer connects the dot suddenly,

we have a lower quieter sound than what was originally recorded. You might

wonder at this point why we don't just sample all of the sound and that is

because at some point we have to pick finite points in time to record

because we do not have unlimited storage space on our computers.

The human ear can no longer distinguish increased fidelity beyond a sample

rate of 48khz so why do we have higher sample rates? The spaces between each

recorded sample essentially becomes instances of noise when the pitch is

recreated. A higher sample rate technically has a better signal to noise

ratio at the cost of storage space. The second reason is that increased

recorded data leads to better mathematical processing when you run your

sound through processing in the form of digital plugins. Small amounts of

inaccuracies can add up overtime when processing takes place leading to

noticeable miss-approximations which we hear as a sound difference.

Differences in bitrate from the 8-bit of old videogames up to the advanced

32-bit level some systems are capable of, is a similar situation to choosing

a sample rate in some respects. A higher bit rate records a larger area in

terms of dynamic range. Let's imagine a bit as a db for this example. In 24

bit the fundamental pitch you are recording of a trombone might be at 20

bits in volume fitting within the max of 24 bits. Each fundamental will then

be quieter than that as we go up the harmonic series. As a result, the

trombone retains its original sound. Let's imagine we are now recording in

8-bit. The trombone still takes up 20 bits in volume, but we can only record

8 of it. The next harmonic which is normally say at 10 bits also doesn't

fit. The fundamental and 1st harmonic are now recorded at the same volume

changing the ratio of the fundamental to harmonics and suddenly the

characteristic sound recorded is no longer the same. Recording in 32 bit is

not necessarily better than 24 unless you sound source is louder than

24bits.

If you are still lost that is okay. Ultimately you don't have to understand

these concepts to work with audio. My recommendation is to record

at 24bit 48khz and everything will likely be okay. I personally record most

clients at 24bit 96khz.

If you want to know more consider scheduling a free session with us,

Spencer Miles Spencer Studios 313 W Liberty St, Lancaster, PA 17603

spencerm96@comcast.net

7176348955