Re: 8 vs 16 bit (was Scott G. vs Reality...)

[pubpc1@library.ucla.edu]

nathan@visi.com (Nathan Mates) wrote,
in response to my reply:
>>You may be right in the technical aspect, but you neglect some
>>physiological aspects of the human ear.  I would venture to say that
>>8-bit resolution is very near the limits of human hearing and any
>>thing higher will only make token differences.  At least to my ears.
>
>   Scott Gamon, this is another technical issue where you've taken
>an illogical, indefensible, and WRONG position, only to claim "I
>notice nothing different."

I never said that.  Sure, I notice a difference, but it is slight.
I was responding to the assertion that the because difference between
1-bit samples and 8-bit samples is so great, 16-bit samples must
show the same night and day difference when compared to 8-bit samples.
That's wrong.  For each increase in resolution, you are going to
notice a smaller increase in quality.


>   Showing this posting of yours to the head of the audio dept where I
>work, the response was "Tell him [Scott] he's stupid." There is a vast
>world of difference between 8 and 16 bit samples; most of the audio
>folks at work can usually guess frequency and sampling rate spot-on
>off a second or two's listen with a pair of speakers more powerful
>than the ones Radio Shack throws in the dumpster.

So is this difference between 8-bit and 16-bit samples as great as
between a 1-bit sample and an 8-bit sample?  No.  If you read what I
wrote, that's the point.  Diminishing increases in quality for each
increase in "bits".


>   Why don't you go write to all the major audio equipment
>manufacturers and audio software developers telling them that since 8
>bits is good enough for you, it's good enough for the world, and they
>^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>should all switch down. I bet your post would become the lunchroom
>^^^^^^^^^^^^^^^^^^^^^^
>bulletin board laughingstock for months.

Did I ever say that?  No.


>>Put in another way, the returns of increasing resolution
>>diminish with increasing resolution.  There is still a difference,
>>however slight, between 8 and 16, but between 16 and 24?  That's
>>overkill.  If I weren't expecting it, I wouldn't even be able to
>>pick out 8-bits.
>
>   And good audiophiles can tell the difference between 16 bit CDs and
>the 20+ bit remastering techniques. I don't care if you're tonedeaf
>and unable to tell the difference between a Speak-and-spell and an
>opera off CD, but what you're describing is *subpar* for humans. Most
>people can tell the difference, and you can't. Don't project the
>entire human race in terms of the genetic structure and synapse
>attangements of "Scott Gamon", as they seem to be better than you in
>terms of hearing.

Was I trying to convert humanity?  No.

Did I say I can't hear the difference between 8-bits and 16-bits?  No.
I said the difference is slight and if I weren't looking for it, I
wouldn't notice.


>>It is the same way with vision.
>
>   No it ain't; your comparison is null and void. You're taking fps
>(i.e. sampling rate) in vision and trying to compare it to levels of
>gradation (i.e. bit depth) in vision. Humans can hear 16 bits worth of
>difference in audio. Humans can see ~24 bits (8 each RGB) of
>difference in video. 256 levels of audio doesn't cut it any better
>than 256 levels (~2.4 bits of RGB) of video would cut it.

I'm saying that as you near the limits of human perception, be it
sampling rate _or_ bit depth, you are going to get less of an
increase in subjective quality for each increase in sampling rate/
bit depth.  Is 16-bit sound superior to 8-bit sound, sure.  But
is that increase anywhere as great as between a 1-bit sound and
an 8-bit one?  No way.  Is the difference between a C64 7-color
graphic and a 256 color GIF greater or the difference between a
GIF and a 24-bit JPEG format image?  Of course there is a difference
btween the GIF and th JPEG, but it isn't as dramatic as between
the C64 picture and the GIF, is it?


>   Humans peak out about 24-28 fps visually. For an equivalent
>sampling rate in audio, they peak out at 22Khz and down (depending on
>gender, age, exposure to excessive noise, etc).
>
>   Compare apples to apples, and oranges to oranges if you want to be
>taken seriously in any type of discussion. Especially, don't think
>that reality is wrong when it doesn't agree with your limited
>perceptions of audio differences, WPGS status, DMA speed, or anything
>else.

I fear that loud music has damaged my ability to hear some high
frequencies, but I can still hear the difference between a 1-bit,
8-bit, and 16-bit sample well enough, thank you.

-Scott G.