[MPA] Audio I/O Interface and Other Newbie Questions

George Reis reis at imagingforensics.com
Tue Feb 18 05:32:55 PST 2003


Scott Baldwin wrote on 2/17/03 6:20 PM:

> The general theory is that although CD's are 16/44k, recording at
> higher rates and then downsampling for the final mix gives a better
> result, than if the music was recorded at 16/44 in the first place.
> This is a high subjective theory, with no real objective way to prove
> it, other than by listening.

Well, I'm not an audio expert, but digital imaging has many similarities to
audio, and bit depth is one of them. In images, the bit depth tells you how
many tonal values there can be between the whitest value and the blackest
value.

A black and white image is described by a brightness value, and the bit
depth is exactly equal to the total number of tones. For instance, an 8 bit
grayscale image has 2 to the 8th bits of data, which is 256, which means 256
shades of gray. A color image is described as multiple values - red, green
and blue (RGB), or cyan, magenta, yellow and black (CMYK), etc. So, a color
RGB image that is 24 bits, actually has 8 bits each of red, green and blue,
and the total description of 24 bits is 2 to the 24th, or 16.7 million
colors, which is the total number of colors that can be described by all the
combinations of 256 shades of red, green, and blue.

Now, suppose that I have a grayscale image that was captured at 8 bits. And,
suppose that I make some modifications to that images - say, increase the
contrast, or adjust the overall brightness. By doing this, I change the
values of some tones, and there frequently would be fewer tones in my new
image - although the overall image may be more generally pleasing. If my
modifications are substantial, the tonal loss can be significant. My final
image will have gaps in the tonal scale, subtle tonal nuances can be easily
destroyed.

Now, suppose that I captured that original image in 16 bit (2 to the 16th,
65,500 tones), and made the same adjustments. I could then downsample that
to 8 bit and eliminate the loss that processing in 8 bit would have given
me. Now I have my improved image, without a loss in the subtle tones.

The same is true in audio.

George
--
George Reis
Imaging Forensics
Cell: 714-315-5161
<http://www.imagingforensics.com>

Specializing in digital imaging consulting, training, analysis and
enhancement for the law enforcement, investigative and legal communities.



More information about the MacProAudio mailing list