In article , joe
wrote:

D. Peter Maus wrote:

On 07/02/09 15:48, Brenda Ann wrote:

2) There is NO mp3 player that can as accurately reproduce a complex
audio
waveform as well as a high end cassette machine. I don't care how many
samples you take of a complex waveform with an ADC/DAC system, the
resultant playback waveform will never represent the original analog
waveform as well as a high end analog device. Even a simple 1000 Hz sine
wave will not come out as a pure sine wave after digital conversion, it
will be a series of stepped square waves. You may not be able to tell the
difference with your ear, as long as there are enough of those little
steps, but that's not the point. The point is, it will not "run circles
around" a high end analog device.





If you take a look at a 1khz square wave after digital
conversion, you'll see ringing at both ends of the flat top. You'll
see that same ringing wherever there is a hard rise or fall. Is it
audible? Oh yeah. More so on a naked square wave. Less so in complex
music. But you can hear it.

You'll see this wherever there is hard digital filtering, such as
anti-aliasing on CD players. You'll see it where there is copious
amount of data loss, as in MP3.

An MP3, at it's best is a 4:1 data loss. The songs on iTunes and
elsewhere are mostly 10:1 data loss. Sometimes more, sometimes less.


Isn't equating compression ration or data rates to data loss a bit
misleading? Sure, MP3 and AAC are lossy codecs, but a lossless codec such
as FLAC reduces the data rate without loss of the original content. The
amount of original signal lost by the use of AAC or MP3 compression is much
less than you imply with your numbers.

Noise may be reduced, but it's hardly high fidelity audio. And
though cassettes have their many flaws, a properly set up Nak will
have more noise, but far less digital artifacting and zero data loss
than any MP3.

I would expect an analog system to have no digital artifacts. But, you admit
there is more noise, isn't that also a loss of 'data'. But you also ignore
any reduction in bandwidth that occurs with magnetic recording. Also, at
its best the Nak may have higher distortion than a high end MP3 player.

Cassette decks have their BW specified at -20 db because at higher levels,
head/media saturation limit useful bandwidth.

There are several technical criteria that must be met with digital
recording so that it can equal analog. These criteria must be met in
both directions, analog to digital, and then digital back to analog
being a complete process.

The sampling rate must be twice the highest frequency you want to record
so if the analog frequency is 22 KHz then the minimum sample rate is 44
KHz. A higher sample rate is better. For this sampling scheme to work
well the conversion in either direction should be low pass filtered.
This scheme has the 22 KHz sine wave represented by two steps, which is
very coarse.

The analog filtering will help the reproduced analog look like the
original recorded analog signal but the conversion sampling has several
types of imprecision to contend with besides sample rate. Just as
important are sample levels. The smaller the sample level the more
precise the reproduced analog will be.

So the two main parameters are the number of samples made in time and
voltage or to look on a sine wave on a graph the horizontal and vertical
axis. The smaller the steps in either axis the closer the digital
stepped waveform approximate the analog. Then a low pass filter smoothes
out the tiny steps as a way of "polishing" the digital waveform to look
even more like the analog.

The problem with the above conversion scheme is the sample imprecision
in time and voltage, which leads to conversion noise and distortion. The
precision can be improved with an increased number of steps in either
axis. Increasing the number of voltage steps means the sampling number
must be numerically larger and increasing the sampling rate increases
the number of samples that must be processed and stored for the same
length of the recording so higher quality means bigger numbers and more
of them.

This is a big problem for digital recording, storage, and reproduction.
If you want high quality you need the electronics to operate rapidly and
generate more data requiring larger storage. The electronics operating
rapidly consumes power and large data storage also costs more money so
the solution is low sample rates and small sample numbers.

Along with small sample numbers and low sample rates, data storage
requirements are further reduced with compression algorithms that are
lossy or in other words further distort the data. Here "lossy" means
some of the data is thrown out and not saved to storage.

The financial cost of these problems also burdens the transmission of
digital data similar for the digital storage cost. Higher quality means
higher transmission rates similar to larger storage requirements. Higher
transmission rates means the signal must occupy more spectrum. For this
reason and others the IBOC and DRM sounding "better" in the same band
space is just plain BS.

So basically, regardless of the sample size and rate used you have
inherent sampling and anti-alias filtering distortion so the converted
analog waveform can never be as good as the original but using more
power, band space, and storage it can be close.

We are all used to the continual improvement in electronics where they
run faster with less power and storage becoming cheaper, smaller, and
lighter with time, so with time all this can be overcome except the
amount of band space needed for transmission. Here improvements in
electronics cannot overcome basic physics.

--
Telamon
Ventura, California