I have been cleaning up a lot of downloads lately because of the file
size. One of the things that puzzle me is the ridiculously high bit
rates that are often chosen for voice only recordings. I convert many
down but that is not the point. The point is that I think most folks
are trying for better quality rips but shooting themselves in the foot
with the choices they make.
There are two major factors that control reproduction. Sample rate
and total bit count. The sample rate sets the band pass. Multiply
the sample rate times the bit count to get the range. In other words
if I sample audio @ 8 KHz it takes two samples to set the highest
frequency or 4 KHz. If I use 4 bits per sample then 4 x 8K = 32K
samples per second. The number of bits sets the dynamic range. 4
bits = 16 discrete levels of sound. CODECS use various schemes to
slide that window around on the volume range of what is being
converted. If you really want better sound increase the rate so you
get more bits per sample.
Here is a bit about why and some easy experiments:
Here is a brief reference to voice channel frequency response used by
communications companies:
https://en.wikipedia.org/wiki/Voice_frequency
As you can see the communications folks figured out that they can do
decent sounding spoken word in 300-3500 Hz bandwidth. They use very
sharp filters when they multiplex voice over air, fiber or wire. If
they do not they get cross talk and other technical problems. They
get away with it because there is nothing there in the higher
frequencies when we are talking about voice. Music is a different
animal.
How good does your old land line phone sound compared to the same
person in person? That is the older phones not the Chinese junk that
may or may not have tailored responses.
If you want another, possibly easier, example tune in a few talk show
hosts on AM radio and again on Television or FM. See if you can tell
the difference on the same set of speakers or headphones. The issue
is AM radio is limited to around 5 KHz band pass while FM and
Television are around 20 KHz. Try the same thing with a concert and
you will see why music moved off AM.
Over the years a lot of very smart people have run all kinds of
experiments proving what works and what does not and why. Please heed
what they learned. Enough band pass to get the frequency response
then worry about dynamic range. We can hear a much broader range of
levels than we can frequencies.
---
This email has been checked for viruses by AVG.
http://www.avg.com
|
|