Questions about equivalents of audio/video and digital/analog.

[Jerry Avins]

I don't want to separate the wing-flapping or anything from the video.
I want all temporal components of the video signal to be slowed
without changing the length or speed of the video. Just like Adobe
Audition and certain voice-changers allow the pitch of the audio
signal to be decreased without changing the length or speed of the
audio.

He wants to walk the same distance at half speed and complete the trip
in the same time as before. I give up!

Jerry

 

[Floyd L. Davidson]

Apparently analog but actually digital? That would be in
keeping with your assertion that quantizing an otherwise
analog signal digitizes it.

You *didn't* quantize it. Or at least nothing you said
assures that it has been quantized, and given "levels can
be individually adjusted" is high suggests that it is not
quantized.

By old vacuum-tube signal generator was certainly
analog. It produced square waves among other wave shapes.

Hot damned, you *are* aware of that. Amazing...

 

[Floyd L. Davidson]

http://www.warwick.ac.uk/~phsbm/qhe.htm

Why don't you tell us what you think this means. I
have no need to spend my time tracking down your
comments.

 

[Ron N.]

I want all temporal components of the video signal to be slowed
without changing the length or speed of the video. Just like Adobe
Audition and certain voice-changers allow the pitch of the audio
signal to be decreased without changing the length or speed of the
audio.

But pitch changing software does change the speed of
all the details in the audio. Instead of 200 vibrations
per phoneme, you might get only 100 after the pitch
changer does its distortion. The onset of the phoneme
will be 5 mS less accurate. I suppose the equivalent
in the temporal component of a video scan line would
be taking a picture of a house with 6 or 8 windows and
changing the house to be about the same width but only
having 3 or 4 windows across, and maybe moving the house
so that it is an even number of window widths from the
house next door. An artist could probably do this kind
kind of caricature for you before storing a lower
data rate description of the resulting picture.

You could also shift the all colors down into infrared,
but couldn't see the result anymore, although you
might be able to use it as a hand warmer on a cold day.

And just how do you that this reply was not generated
somehow involving an escaped AI experiment?

 

[Don Bowey]

You don't appear to understand that the limited set of values makes
it digital, by definition. PERIOD.

Of course if you then run that digital PAM signal through virtually
any analog channel, it no longer has a limited set of values...

Including a two foot piece of cable, or two inches with a small cap.

 

[glen herrmannsfeldt]

Radium wrote:

(snip)

Real-time pitch-shifting is done for audio on the phone. Certain voice-
changing devices allow women to sound like men on the phone, without
decreasing the speed at which they talk. The pitch of the audio is
decreased but the speed remains the same.

I would like something similar to be done with the spatial and
temporal of a video signal in real-time. I would like to be able to
work not only for recorded video but also for video signals that are
being transmitted/received in real-time -- such as a live TV show.

Normally this is done by removing the appropriate frames (samples),
or duplicating them. One example is the 3-2 pulldown used to convert
24 frame/second movies to 59.94 field/second video.

Actually I don't want other parts to be unchanged. What I would like
is the temporal frequencies [of all parts of the video] to be
decreased but without decreasing the speed of the video signal.

That might be harder. For audio, it can be chopped, such as
removing 1/60th of a second every 1/20th to speed it up by 1.5.
If that is done faster than the modulation (vocalization),
and slower than the frequencies of interest (maybe 400Hz-3.5kHz
for voice) it works pretty well. Probably less well for music.

As far as I know, that is usually done asynchronously to the
source signal. One could imagine removing cycles of a 1.23kHz
voice, for example.

For video the modulation (wing flapping) is not separate from
the source frequency. If you know you are trying to separate
wing flapping, it could be done by cutting out whole flap
cycles, assuming only one bird is in the scene, and is doing
most of the motion. Otherwise, I don't think there is anything
you could do.

-- glen

 

[glen herrmannsfeldt]

Floyd L. Davidson wrote:
(snip)

Actually I suspect it is open to debate as to whether a
sample is actually PAM until it is quantized. (Until it
is, it's just a sample of an analog signal.) But
whatever, if the sample itself actually is PAM, then yes
that is an analog signal.

Phase modulation can be done continuously, and so can
amplitude modulation. The NTSC chroma subcarrier is
pretty much PAM, where the phase indicates the hue and
amplitude the intensity of the color.

NTSC isn't actually described that way due to the desire to get
just a little more resolution out of the signal. The eye is
more sensitive to spatial resolution between some colors than
others (and not along the obvious R-G-B axes), and NTSC encodes
that. Only very recently were TV receivers built to decode that
extra information, at about the same time ATSC tuners are taking
over.

-- glen

 

[Radium]

But pitch changing software does change the speed of
all the details in the audio. Instead of 200 vibrations
per phoneme, you might get only 100 after the pitch
changer does its distortion. The onset of the phoneme
will be 5 mS less accurate.

I am fine with that.

I suppose the equivalent
in the temporal component of a video scan line would
be taking a picture of a house with 6 or 8 windows and
changing the house to be about the same width but only
having 3 or 4 windows across, and maybe moving the house
so that it is an even number of window widths from the
house next door. An artist could probably do this kind
kind of caricature for you before storing a lower
data rate description of the resulting picture.

I still think a more precise example of an equivalent is the wing-
flapping I described.

In the video signal containing the flapping-wings, all temporal
components of the video [including the flap-rate] are slowed without
increasing the length of the video. The video clip remains just as
short.

 

[Floyd L. Davidson]

Including a two foot piece of cable, or two inches with a small cap.

Nope. It would take a fair sized cap.

Keep in mind that that is *exactly* what a V.90 modem puts on a
regular twisted pair telephone cable, and it works just fine for
a couple miles at least, sometimes even much farther.

And that signal is digital, and is processed as a digital signal
by the receiving modem.

 

[Floyd L. Davidson]

Floyd L. Davidson wrote:

(snip)



There needs to be a way to describe sampled but not quantized
signals. They are not continuous in time, but the function
can take on any value at each sample point.
Exactly.

It does occur to me that you could have quantized but not
sampled data. It works best gray coded so that there isn't
a problem at transitions. This is sometimes done for rotational
encoders, where the rotation angle can be continuous such
that transitions can occur at any time.

 

[glen herrmannsfeldt]

Floyd L. Davidson wrote:

(snip)

Analog signals are by *definition* continous.

You have misunderstood what that means though. The
analog value of a signal is continuous, but that does
not imply that the signal continuously exists or that
it even changes at all.

There needs to be a way to describe sampled but not quantized
signals. They are not continuous in time, but the function
can take on any value at each sample point.

It does occur to me that you could have quantized but not
sampled data. It works best gray coded so that there isn't
a problem at transitions. This is sometimes done for rotational
encoders, where the rotation angle can be continuous such
that transitions can occur at any time.

-- glen

 

[isw]

Hi:

I. Audio vs. Video

Digitized (mono) audio has a single sample per each sampling
interval.

Yes. several bits per sample, many samples per second.

In the case of digital video, we could treat each individual sample
point location in the sampling grid (each pixel position in a frame)
the same way as if it was a sample from an individual (mono) audio
signal that continues on the same position in the next frame. For
example, a 640?480 pixel video stream shot at 30 fps would be treated
mathematically as if it consisted of 307200 parallel, individual mono
audio streams [channels] at a 30 Hz sample rate. Where does bit-
resolution enter the equation?

It might actually make sense to look at it that way in some situations,
but I'll bet you can't think of one.

How about a T1 (DS1) stream? It's a series of 8-bit audio samples, with
frame sync.

Isaac

 

[Don Bowey]

Nope. It would take a fair sized cap.

Keep in mind that that is *exactly* what a V.90 modem puts on a
regular twisted pair telephone cable, and it works just fine for
a couple miles at least, sometimes even much farther.

And that signal is digital, and is processed as a digital signal
by the receiving modem.

Digital data CSUs and T1 transmitter line signals are digital and look
similar to distorted square waves. An all 1's signal looks like a distorted
sinewave .

Using the same V.90 example....... It will work as well if two v.90 modems
are connected back-to-back by a short pair of wires.

 

[Jerry Avins]

You *didn't* quantize it. Or at least nothing you said
assures that it has been quantized, and given "levels can
be individually adjusted" is high suggests that it is not
quantized.


Hot damned, you *are* aware of that. Amazing...

You know, Floyd, legal definitions don't always reflect reality. Back in
the 50s, the speed limit on Storrow drive in Boston was 35 mph. By law,
exceeding 50 mph in a 35 mph zone was statutory reckless driving. To
expedite traffic, the police let it be known publicly that they wouldn't
give speeding tickets during rush-hour, leaving motorists free to travel
up to 50 mph without consequence. One night about 1:00 AM I got a ticket
for reckless driving by going 55 mph on an empty road. I pleaded not
guilty and offered to plead guilty to speeding. I argued that if 50 mph
was perfectly OK on a crowded road, 55 mph wasn't reckless no matter how
the law read. I remember the judges words: "You're a little snotnose.
I'd love to fine you for speeding, but you're not charged with that.
Case dismissed!"

I'm still a little snotnose, and partitioning a signal into approximate
levels -- because of noise the levels can never be exact -- doesn't make
it digital *in fact*, whatever the gummint might declare.

Jerry

 

[Ron N.]

I don't want to separate the wing-flapping or anything from the video.
I want all temporal components of the video signal to be slowed
without changing the length or speed of the video. Just like Adobe
Audition and certain voice-changers allow the pitch of the audio
signal to be decreased without changing the length or speed of the
audio.

Pitch shifters remove (or duplicate) cycles. For instance,
if you have a video of a car driving past a house with six
windows in one second, and you slow the video down to half
speed it will normally take two seconds to show the complete
video. However, if you just chop out of the video portions
showing the car between, say, the odd numbered windows of the
house, then you can show the jerky left over video in only
one second. If you chopped out the odd numbered windows
from the pictures of the house, and stretched out the portions
with the even numbered windows, then it might look like a
car twice as long was driving in front of a house with less
windows, or it might look like something Dali or Picasso
would draw during a hangover.

 

[Jerry Avins]

Why don't you tell us what you think this means. I
have no need to spend my time tracking down your
comments.

Live and learn, or live and don't learn. Your choice. I'm indifferent.

Jerry

 

[Jerry Avins]

Nope. It would take a fair sized cap.

Keep in mind that that is *exactly* what a V.90 modem puts on a
regular twisted pair telephone cable, and it works just fine for
a couple miles at least, sometimes even much farther.

And that signal is digital, and is processed as a digital signal
by the receiving modem.

The signal is amplified in analog repeaters and again in the modem. Your
categories are too hard edged. The very purpose of a modem is converting
digital signal to analog that can traverse an analog phone line and back
again to digital at the far end. A modem might be the worst example of a
purely digital I can imagine.

Jerry

 

[Jerry Avins]

Hi:

I. Audio vs. Video

Digitized (mono) audio has a single sample per each sampling
interval.

Yes. several bits per sample, many samples per second.

In the case of digital video, we could treat each individual sample
point location in the sampling grid (each pixel position in a frame)
the same way as if it was a sample from an individual (mono) audio
signal that continues on the same position in the next frame. For
example, a 640?480 pixel video stream shot at 30 fps would be treated
mathematically as if it consisted of 307200 parallel, individual mono
audio streams [channels] at a 30 Hz sample rate. Where does bit-
resolution enter the equation?

It might actually make sense to look at it that way in some situations,
but I'll bet you can't think of one.

How about a T1 (DS1) stream? It's a series of 8-bit audio samples, with
frame sync.

Isaac

Are you Radium in disguise? My bet was with him.

Jerry

 

[Floyd L. Davidson]

Digital data CSUs and T1 transmitter line signals are digital and look
similar to distorted square waves. An all 1's signal looks like a distorted
sinewave .

Your point is? (Besides the poor description? They
don't look like distorted square waves. The look like
only slightly distorted sine waves!)

Using the same V.90 example....... It will work as well if two v.90 modems
are connected back-to-back by a short pair of wires.

It won't. They can't talk to each other that way except
using v.34 protocols.

Regardless, what is your point? I said that v.90 works
fine for a couple of *miles*, minimum, so what
significance would there be to working "back-to-back by
a short pair of wires"?

 

[Floyd L. Davidson]

You know, Floyd, legal definitions don't always reflect
reality.

You know Jerry, *technical* definitions are reality.

You can fight it all you like, but it won't change the
fact that to talk to anyone about this topic *requires* that
we all use the same definitions, and the ones that I've
cited *are* the standard definitions used by *everyone*
that is credible.

....

law read. I remember the judges words: "You're a little
snotnose. I'd love to fine you for speeding, but you're
not charged with that. Case dismissed!"

I'm still a little snotnose, and partitioning a signal
into approximate levels -- because of noise the levels
can never be exact -- doesn't make it digital *in fact*,
whatever the gummint might declare.

You aren't a little snotnose. Your just a little
foolish, that's all. Everyone is about something, and
that's where you've chosen to make your stand.

Regardless of how silly you want to be, it *does* make
it digital, by definition.