"Mike Rivers" wrote
Yeah, I guess that most pop music is mostly mono. They
don't want to take a chance that the driver of the boom-SUV will miss
anything that goes on over on the passenger's side, and vice versa.

Not at all. It has to do with in the 60's they used to separate things
really wide. John would be in the left speaker only and Paul only in the
right. The drums in one speaker and the guitar in the other. Those were
weird mixes and don't sound anything like reality. Engineers figured out
reality is mostly mono, so they started mixing mostly mono to make things
sound more realistic. There is an art to making believable stereo mixes.
They are mostly mono, but certain things, like reverbs, delays etc., ARE
spaced as far apart as possible. Its just that what sounds the most real is
usually about 80% mono. REALLY!

I'm starting to get the impression that the end point here isn't to
restore the original stereo spread and image, but rather to make the
stereo (as much of it as there is) into mono, then bugger it so that
there's some stereo spread - which doesn't necessarily have to be the
same as what went in to the encoder.

No, you misunderstand. L+R combined with L-R is EXACTLY the same as L and
R. Nothing is buggered. Nothing is lost, nothing is added. Now encoding
is another story and stuff IS lost and added, but none of the lost and added
is due to the L+R / L-R part.

In FM radio they do it so there is an easy mono signal for places where the
reception is too weak for stereo. In MPG they do it because it improves the
quality.

Julian

On 6 Apr, in article znr1112790419k@trad
"Mike Rivers" wrote:

In article writes:


Mike Rivers wrote:
However, I really have no idea just what this L+R/L-R recording method
really is, how it works, why it works, or what's behind it all.

Basically: Rather than two cardioid microphone cartridges pointing
across each other to record your stereo image, use one cardioid pointed
straight ahead and a bidirectional at right angles to it.

Right - I know all about that. I want to know what the benefit of
converting left and right inputs to sum and difference before dropping
out data that's subjectively insignificant is. And why it's still
insignificant when converted back to left and right. But I probably
wouldn't understand the answer. Maybe there's an article in the AES
Journal that I haven't got around to reading yet.

Isn't most lossless stereo a bit of a fake anyway?

Not the live recording with two mics, but the studio work -- as I recall
it, the individual tracks are essentially mono, mixed together, and that
mixing to create the stereo image only uses volume differences, not the
phase differences that would be there in a true binaural recording. And
that makes the difference signal a lot simpler.

Does that make any sort of sense, or am I badly out of date on what
happens in a studio?

--
David G. Bell -- SF Fan, Filker, and Punslinger.

"I am Number Two," said Penfold. "You are Number Six."

""David G. Bell"" wrote

Isn't most lossless stereo a bit of a fake anyway?

Not the live recording with two mics, but the studio work -- as I recall
it, the individual tracks are essentially mono, mixed together, and that
mixing to create the stereo image only uses volume differences, not the
phase differences that would be there in a true binaural recording. And
that makes the difference signal a lot simpler.

Does that make any sort of sense, or am I badly out of date on what
happens in a studio?

I'm badly out of date in a studio myself. I last mixed an album a couple
years ago and its been 5 or 6 years before that since I did it regularly.

But, yes, it is a fake, but an elaborate one Good engineers who use good
mics, good mic placement, stereo mics, delays, eq effects, and reverbs can
create amazing images. Yes they're fakes, but if you close your eyes its
like being there. I've done a lot of live recording too, which actually is
done being there and the fake studio stuff can be every bit as rich.

Julian

In article writes:

No, you misunderstand. L+R combined with L-R is EXACTLY the same as L and
R. Nothing is buggered. Nothing is lost, nothing is added. Now encoding
is another story and stuff IS lost and added, but none of the lost and added
is due to the L+R / L-R part.

I understand all of that. I've used the mic technique for years. What
I don't understand is why they do this when encoding MP3. Someone
started to give an explanation that mono encodes more efficiently, and
that's what L+R is. I'll accept that for now.




--
I'm really Mike Rivers - )
However, until the spam goes away or Hell freezes over,
lots of IP addresses are blocked from this system. If
you e-mail me and it bounces, use your secret decoder ring
and reach me he double-m-eleven-double-zero at yahoo

"Mike Rivers" wrote

I understand all of that. I've used the mic technique for years. What
I don't understand is why they do this when encoding MP3. Someone
started to give an explanation that mono encodes more efficiently, and
that's what L+R is. I'll accept that for now.

Hi Mike,

I'll try one last time in over simplified terms. IF 80 % (roughly) of the
information is MONO, THEN only 20% is L-R. It takes a lot less data to
encode the small amount of 20% L-R information than either the full
bandwidth L or R or Mono channels.

Julian

Encoding L+R and L-R is more efficient then encoding L and R because
the L+R carries most of the information so you have one "major"
channel to encode and one minor (the L-R) channel instead of 2 major
channels.

Mark

nicely said

"Mark" wrote in message
ups.com...
Encoding L+R and L-R is more efficient then encoding L and R because
the L+R carries most of the information so you have one "major"
channel to encode and one minor (the L-R) channel instead of 2 major
channels.

Mark

Julian Adamaitis wrote:
nicely said


thank you

Mark

In article writes:

I'll try one last time in over simplified terms. IF 80 % (roughly) of the
information is MONO, THEN only 20% is L-R. It takes a lot less data to
encode the small amount of 20% L-R information than either the full
bandwidth L or R or Mono channels.

I understand that. Oversimplification of a very complex process
doesn't work for me.

If there's a useful oversimplification, it's the assumption that 80%
of the information is mono. How is this deduced? Surely you don't have
80% complete duplication in the two channels. Maybe you have 80%
duplication if you allow a 20% (or some other figure) fudge factor,
saying that the two channels are 'close enough' 80% of the time. Is
that how it works?

This may be valid for a pop recording, but on an at least somewhat
professional recorder like the Marantz (which, sadly, seems to have
some quite un-pro features) you'd think they'd want to do better. I
doubt that true two-mic stereo recordings have near 80% mono content.
But then I've never thought about it, and I don't really know how to
go about thinking about it. But I've seem plenty of Lissajous patterns
and few of them look like tight ovals.




--
I'm really Mike Rivers - )
However, until the spam goes away or Hell freezes over,
lots of IP addresses are blocked from this system. If
you e-mail me and it bounces, use your secret decoder ring
and reach me he double-m-eleven-double-zero at yahoo

"Mike Rivers"

I understand that. Oversimplification of a very complex process
doesn't work for me.

Try Mike's simplified version in this thread! I think maybe even you can
accept it! It is absolutely true but doesn't get caught in concepts that
require theory.

If there's a useful oversimplification, it's the assumption that 80%
of the information is mono. How is this deduced? Surely you don't have
80% complete duplication in the two channels. Maybe you have 80%
duplication if you allow a 20% (or some other figure) fudge factor,
saying that the two channels are 'close enough' 80% of the time. Is
that how it works?

No I'd GUESS its typically 80%, but it can vary drastically with program
material. Like I said earlier the old Beatles stuff where everything was
panned hard left and right was most certainly much less than 80% mono. I
thought I made it clear the exact number was not necessary to understand the
concept of L-R encoding. What is important is that IF *L-R is less than
L+R* you save data. Even of you have 60/40% the techniques still works

Why did I say 80%? I WAS GUESSING a purely ballpark number based on albums
I've personally mixed and albums I've watched excellent engineers mix. I
pan most instruments dead center or 10 /11 o'clock left or 1/2 o'clock
right. Few things I pan 3:00 / 9:00 like stereo drum overheads, stereo
pianos etc., and those things usually have much common information so there
still isn't that much difference between left and right channels. The ONE
and ONLY thing that do I always pan hard left and right is reverb, which in
total volume is the quietest part of the entire mix.

Based on that I came up with a totally seat of the pants number 80% mono.
The technique works if it is only 60% mono. You are welcome to come up with
a more precise number if that's what your after. My only point is it sounds
better the more difference there is. Your criticism of my non-technical
explanation is inappropriate as I was responding to a guy who just wasn't
getting even after reading 3 explanations and being MORE precise would have
probably confused him even MORE. You are welcome to come up with you own
explanation that is both technically precise and simple to understand to non
technical people. I look forward to reading it if you do so!

This may be valid for a pop recording, but on an at least somewhat
professional recorder like the Marantz (which, sadly, seems to have
some quite un-pro features) you'd think they'd want to do better. I
doubt that true two-mic stereo recordings have near 80% mono content.
But then I've never thought about it, and I don't really know how to
go about thinking about it. But I've seem plenty of Lissajous patterns
and few of them look like tight ovals.

The small amount of stuff I mix that IS panned very hard is VERY much out of
phase That's the whole point of spreading it out as far as possible! It's
still a minority of total decibels however. It still takes significantly
less data to describe the out of phase material than the in phases material.

Julian

Mark wrote:
Julian Adamaitis wrote:
nicely said


thank you

Mark

by the way...

I think I read he

http://harmsy.freeuk.com/mosty=ADnc/

that the MP3 encoder automatically switches back to L and R encoding if
it sees that the L-R signal is too complex and the L+R and L-R encoding
stratagy would fail to give better results.
So you get the best of both worlds, if there is a benefit to the L+R
L-R encoding, it uses it, if not, it doesn't.

Mark

"Mark" wrote

by the way...

I think I read he

http://harmsy.freeuk.com/mosty*nc/

that the MP3 encoder automatically switches back to L and R encoding if
it sees that the L-R signal is too complex and the L+R and L-R encoding
stratagy would fail to give better results.
So you get the best of both worlds, if there is a benefit to the L+R
L-R encoding, it uses it, if not, it doesn't.

Mark

That makes sense. If the mono signal is more than difference, it would
actually take more data to do it sum difference. Seeing as popel can create
anything from mono to mostly stereo, a system that didn't know what to do in
that case would be of limited use.

Julian

(Mike Rivers) wrote in news:znr1112958535k@trad:

If there's a useful oversimplification, it's the assumption that 80%
of the information is mono. How is this deduced? Surely you don't have
80% complete duplication in the two channels. Maybe you have 80%
duplication if you allow a 20% (or some other figure) fudge factor,
saying that the two channels are 'close enough' 80% of the time. Is
that how it works?

It's a reasonable assumption. Typically the most important source is
centered. That applies to classical and modern content alike. If you
encode the "mid" with high res and the "side" with low res, the most you
can lose is the stereo image.

Carey Carlan wrote:
(Mike Rivers) wrote in news:znr1112958535k@trad:

If there's a useful oversimplification, it's the assumption that 80%
of the information is mono. How is this deduced? Surely you don't have
80% complete duplication in the two channels. Maybe you have 80%
duplication if you allow a 20% (or some other figure) fudge factor,
saying that the two channels are 'close enough' 80% of the time. Is
that how it works?


It's a reasonable assumption. Typically the most important source is
centered. That applies to classical and modern content alike. If you
encode the "mid" with high res and the "side" with low res, the most you
can lose is the stereo image.

Phase anomalies between the two recorded tracks were quite common in the
analog days, especially after the signal had passed through a couple
generations and at least one pass through a routing switcher. M-S
encoding drastically reduced the sonic impact of those anomalies (while
preserving mono compatibility.)

In article writes:

The small amount of stuff I mix that IS panned very hard is VERY much out of
phase That's the whole point of spreading it out as far as possible! It's
still a minority of total decibels however. It still takes significantly
less data to describe the out of phase material than the in phases material.

I hope I'm misunderstand you here, but I hope you aren't purposely
making recordings that I would consider unlistenable just for the sake
of making the compression algorithm work better. Say it aint' so, Joe.




--
I'm really Mike Rivers - )
However, until the spam goes away or Hell freezes over,
lots of IP addresses are blocked from this system. If
you e-mail me and it bounces, use your secret decoder ring
and reach me he double-m-eleven-double-zero at yahoo

"Mike Rivers" wrote

I hope I'm misunderstand you here, but I hope you aren't purposely
making recordings that I would consider unlistenable just for the sake
of making the compression algorithm work better. Say it aint' so, Joe.

You are misunderstanding me. My goal is to make natural sounding
recordings. Having Ringo sing out of one speaker and Paul out of the other
is not making a natural sounding recording IMO! Mike you seem very
knowledgeable on a lot of subjects. I don't understand why you think
"listenable" recordings means stuff panned hard???? Simply not true, my
good man!

Julian

In article writes:

Mike you seem very
knowledgeable on a lot of subjects. I don't understand why you think
"listenable" recordings means stuff panned hard???? Simply not true, my
good man!

I don't. But what I read from your post to which I commented, I
thought you said you intentionally recorded with wide separation.

Forget it. I saw a Marantz 660 over the weekend anyway.

--
I'm really Mike Rivers )
However, until the spam goes away or Hell freezes over,
lots of IP addresses are blocked from this system. If
you e-mail me and it bounces, use your secret decoder ring
and reach me he double-m-eleven-double-zero at yahoo

Anyone out there with a PMD660? I have one, and I wonder if you have
the same problem I do:

With no pad on, levels are normal. Unit is overdriven with moderate
inputs, faster than I would like, but this isn't the problem. When I
switch on the 20db pad, the level decreases by 38db instead of 20db. I
checked this with a repeatable sound source (test tone through computer
speakers) and got consistent results with different mics and phantom
power on/off.

I believe "20db pad" means the peaks should be 20db lower, right?
Denon/Marantz has taken a while to answer this question, so I'm posing
it to the RAP community.

It makes this unit nearly unusable because 0db pad distorts amazingly
fast, and 20db pad (actually 38db pad) requires turning up the gain
until the thing is pretty noisy. That, and the fact that I can't see
the LED meters in bright sunlight makes it difficult to make field
recordings of a loud, outdoor activity (drum & bugle corps).

Eric
--- change x to z to reply

"Mike Rivers" wrote

I don't. But what I read from your post to which I commented, I
thought you said you intentionally recorded with wide separation.

No, I said the exact opposite.

Julian

In article 1113361704.4cade86c7b197c5e55469b302c428e3a@teran ews writes:

With no pad on, levels are normal. Unit is overdriven with moderate
inputs, faster than I would like, but this isn't the problem. When I
switch on the 20db pad, the level decreases by 38db instead of 20db. I
checked this with a repeatable sound source (test tone through computer
speakers) and got consistent results with different mics and phantom
power on/off.

I believe "20db pad" means the peaks should be 20db lower, right?

While it might be the peaks that count, it's difficult to measure
attenuation using peak values. This is where a good old fashioned VU
(or VU-style) meter is useful. I've only fondled one of these
recorders and not had an opportunity to put it on the bench, but it's
easy enough to make a 20 dB pad that I can't imaging that they got
that wrong - unless it's not really a pad. That would be bad.

It makes this unit nearly unusable because 0db pad distorts amazingly
fast, and 20db pad (actually 38db pad) requires turning up the gain
until the thing is pretty noisy. That, and the fact that I can't see
the LED meters in bright sunlight makes it difficult to make field
recordings of a loud, outdoor activity (drum & bugle corps).

Some people have it in mind that the recording level is too low if the
meters don't hit the peak much of the time. Perhaps you should try
recording with the pad in and just turn up your playback volume a bit.
Is the noise you get when you turn up the gain real electronic noise,
or are you talking about background noise? Of course that will come up
(along with what you really want to record) when you turn up the gain.

One of the questions I asked of the person whos PMD660 I was looing at
over the weekend was whether the meters were easy to read. He said
they were, so I guess we have a difference of opinion there. He mostly
records acoustic instruments in jam sessions, so overload is no
problem for his application. In fact he was pretty happy using the
internal microphones (which he was using when I saw him with the
recorder).



--
I'm really Mike Rivers )
However, until the spam goes away or Hell freezes over,
lots of IP addresses are blocked from this system. If
you e-mail me and it bounces, use your secret decoder ring
and reach me he double-m-eleven-double-zero at yahoo

Thanks for your response! It is very strange the way this thing PMD660
pad operates, and this is the first recorder I have with a pad setting
so I hope it's not user error. With the pad engaged the levels are
microscopic, so it appears something is wrong.

In my first recording, I could not use the no-pad because I get
distortion even with moderate volumes. Changing the gain knob merely
sets what level the flattops occur! With the pad in, I turn the gain
knob up to 75% or more and the levels are still peaking at -30 or so. I
did turn up playback volume (or normalized) but this made my recording
sound like it was done oceanside!

Although far from pro, I've done enough recordings to know something
funky is going on. It is making my Sharp minidisc look pretty good.
Maybe another PMD660 owner encountered this problem since I noticed it
within the first few minutes of use.

For the meters, I can probably build a hood or wear the recorder around
my neck.

Eric



In article znr1113391513k@trad, says...

In article 1113361704.4cade86c7b197c5e55469b302c428e3a@teran ews writes:

While it might be the peaks that count, it's difficult to measure
attenuation using peak values. This is where a good old fashioned VU
(or VU-style) meter is useful. I've only fondled one of these
recorders and not had an opportunity to put it on the bench, but it's
easy enough to make a 20 dB pad that I can't imaging that they got
that wrong - unless it's not really a pad. That would be bad.

Some people have it in mind that the recording level is too low if the
meters don't hit the peak much of the time. Perhaps you should try
recording with the pad in and just turn up your playback volume a bit.
Is the noise you get when you turn up the gain real electronic noise,
or are you talking about background noise? Of course that will come up
(along with what you really want to record) when you turn up the gain.

One of the questions I asked of the person whos PMD660 I was looing at
over the weekend was whether the meters were easy to read. He said
they were, so I guess we have a difference of opinion there. He mostly
records acoustic instruments in jam sessions, so overload is no
problem for his application. In fact he was pretty happy using the
internal microphones (which he was using when I saw him with the
recorder).

In article 1113521327.94035ea97bb99a42b737db710f5360f8@teran ews writes:

In my first recording, I could not use the no-pad because I get
distortion even with moderate volumes. Changing the gain knob merely
sets what level the flattops occur!

That means that the mic is clipping the front end of the preamp, ahead
of the pad (bad) and ahead of the gain/record level control.

With the pad in, I turn the gain
knob up to 75% or more and the levels are still peaking at -30 or so.

Maybe it's broken, maybe it's a bug. Can you generate a steady tone
that you can feed to the mic input and actually see what the pad is
doing?


--
I'm really Mike Rivers )
However, until the spam goes away or Hell freezes over,
lots of IP addresses are blocked from this system. If
you e-mail me and it bounces, use your secret decoder ring
and reach me he double-m-eleven-double-zero at yahoo

Recall that I have a Marantz PMD660 where the 20db pad seemed to remove
more than 20db of stuff. With the pad in, I turn up the gain to 75%
with a decent source, peaks are still at -30 or so. The pad was simply
too much and this made no sense.

In article znr1113526669k@trad, says...

Maybe it's broken, maybe it's a bug. Can you generate a steady tone
that you can feed to the mic input and actually see what the pad is
doing?

Yes, it was broken or something like this. After speaking with the
people at Denon/Marantz (and sending them audio files) they finally
wrote a return authorization and I mailed it back. They returned it
VERY QUICKLY without explanation, except that the work required four 22K
resistors.

Everything works great, gain staging is great, levels are decent, no
distortion. This is a pretty good little recorder except for a few
personal preferences. It's a mystery why my unit supposedly had
incorrect or missing resistors in it. I bet there's more in the same
condition, waiting to be discovered.

Eric
---- change x to z to reply

In article 1115082804.7ccbcd5f1ca70449e04189b568c3f32b@teran ews writes:

Recall that I have a Marantz PMD660 where the 20db pad seemed to remove
more than 20db of stuff. With the pad in, I turn up the gain to 75%
with a decent source, peaks are still at -30 or so. The pad was simply
too much and this made no sense.

After speaking with the
people at Denon/Marantz (and sending them audio files) they finally
wrote a return authorization and I mailed it back. They returned it
VERY QUICKLY without explanation, except that the work required four 22K
resistors.

Everything works great, gain staging is great, levels are decent, no
distortion. This is a pretty good little recorder except for a few
personal preferences. It's a mystery why my unit supposedly had
incorrect or missing resistors in it. I bet there's more in the same
condition, waiting to be discovered.

Often they make mistakes and things have to be patched up. If they
catch it in time, you'll find components or jumpers tacked in to a
brand new unit. If they don't, then they have to fix it later. Of
course it's cheaper to do it right from the beginning, but they know
that a certain number of users will never discover, or will never be
affected by a built-in problem. Good thing you figured out that
something had to be wrong and that there was a solution.

I recall in the early days of the Mackie CR1604 mixer there was an
incorrect value resistor in early production runs and the two meters
didn't read the same for an identical input signal.

--
I'm really Mike Rivers )
However, until the spam goes away or Hell freezes over,
lots of IP addresses are blocked from this system. If
you e-mail me and it bounces, use your secret decoder ring
and reach me he double-m-eleven-double-zero at yahoo