[Carey Carlan]

I just received a CD of a cassette of my great aunt recorded in the 1960's.
I'm still working on getting the original cassette. No guarantees.

It's a stereo recording of a mono signal. I want to capture only the
information common to both channels and discard any differences as noise.

I should be able to do that with an M-S converter, discarding the side
channel entirely.

Is there a better way?

[Federico]

If it is a mono recording my advice is to choose the channel that sound
better and discard the other...
F.

[Scott Dorsey]

Carey Carlan wrote:
I just received a CD of a cassette of my great aunt recorded in the 1960's.
I'm still working on getting the original cassette. No guarantees.

It's a stereo recording of a mono signal. I want to capture only the
information common to both channels and discard any differences as noise.

I should be able to do that with an M-S converter, discarding the side
channel entirely.

The middle channel of the M-S decoder is just the sum of the two channels.
That is, it's the same thing you'd get if you flipped the mono switch.

Is there a better way?

You are probably better off using just one of the two channels instead of
both, because the azimuth on the playback will have been incorrect and the
top end will therefore be degraded when you sum to mono. Just using one
channel gives you more noise, but it gives you some of your high end back.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."

[Carey Carlan]

"Soundhaspriority" wrote in
:


"Carey Carlan" wrote in message
...
I just received a CD of a cassette of my great aunt recorded in the
1960's.
I'm still working on getting the original cassette. No guarantees.

It's a stereo recording of a mono signal. I want to capture only the
information common to both channels and discard any differences as
noise.

I should be able to do that with an M-S converter, discarding the
side channel entirely.

Is there a better way?

I agree with Federico and Scott. The problem is that the tape wiggles
as it moves, making the time relationship between the two channels
inconsistent.

Understood. I'll report back. Background noise levels are about 15 dB
below signal peaks. It's going to be fun...

[Scott Dorsey]

Carey Carlan wrote:
"Soundhaspriority" wrote in
"Carey Carlan" wrote in message
...
I just received a CD of a cassette of my great aunt recorded in the
1960's.
I'm still working on getting the original cassette. No guarantees.

It's a stereo recording of a mono signal. I want to capture only the
information common to both channels and discard any differences as
noise.

I should be able to do that with an M-S converter, discarding the
side channel entirely.

Is there a better way?

I agree with Federico and Scott. The problem is that the tape wiggles
as it moves, making the time relationship between the two channels
inconsistent.

Understood. I'll report back. Background noise levels are about 15 dB
below signal peaks. It's going to be fun...

You really, really want the original.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."

[Federico]

Understood. I'll report back. Background noise levels are about 15 dB
below signal peaks. It's going to be fun...

Do not try to clean the audio too much.
You can get the best results with many different little tweaks, such as eq
+ denoising + eq again.
But, as Scott said, try to get the originals. And aquire them at 24bit, not
16.
F.

[Anahata]

On Mon, 15 Dec 2008 17:55:43 +0000, Carey Carlan wrote:

It's a stereo recording of a mono signal. I want to capture only the
information common to both channels and discard any differences as
noise.

Lots of good advice already given: the simple solution is to sum to mono,
a pragmatic improvement is to pick the best channel and use that, and
often with cassettes there's a big difference between L and R.

A possible further improvement: it's quite likely that there is some
azimuth error in the playback which shows as a timing difference between
left and right channels. If you can apply separately variable time delays
to left and right channels, you could adjust the relative timing before
summing for best (judged by ear) results. The tape will have wobbled, but
at least you can tune out any constant average difference and get some
improvement.

That's still an approximation though: as Scott says, you really want the
original tape and a cassette playback transport which is fully
mechanically adjustable for tape alignment and azimuth.

Digital noise removal works quite well on low-level tape hiss, but if the
noise is only 15dB below the peaks you will have to make a compromise
between residual hiss and noise removal artifacts.

--
Anahata
==//== 01638 720444
http://www.treewind.co.uk ==//== http://www.myspace.com/maryanahata

[Don Pearce[_2_]]

On Mon, 15 Dec 2008 17:55:43 GMT, Carey Carlan
wrote:

I just received a CD of a cassette of my great aunt recorded in the 1960's.
I'm still working on getting the original cassette. No guarantees.

It's a stereo recording of a mono signal. I want to capture only the
information common to both channels and discard any differences as noise.

I should be able to do that with an M-S converter, discarding the side
channel entirely.

Is there a better way?

You've already had a lot of good advice about picking the best channel
and using that, but there is a tool you could try before doing that.
Adobe Audition and its predecessor CoolEdit have a tool called Centre
Channel Extractor. What it does is exactly what you say above. It
examines the stereo signal, amplifies whatever is identical in both
channels and suppresses whatever doesn't correlate - which would be
most of the noise.

It may turn your -15dB noise into -25 or -30dB, which will make a huge
difference when it comes to digital noise removal.

d

[[email protected]]

Carey Carlan wrote:
I just received a CD of a cassette of my great aunt recorded in the 1960's.
I'm still working on getting the original cassette. No guarantees.

It's a stereo recording of a mono signal. I want to capture only the
information common to both channels and discard any differences as noise.

I should be able to do that with an M-S converter, discarding the side
channel entirely.

Is there a better way?

I suggest you post this question at comp.dsp

Mark

[Don Pearce[_2_]]

On Tue, 16 Dec 2008 12:26:55 -0500, "Soundhaspriority"
wrote:


Don Pearce wrote in message news:49477b34.503246718@localhost...
On Mon, 15 Dec 2008 17:55:43 GMT, Carey Carlan
wrote:

I just received a CD of a cassette of my great aunt recorded in the
1960's.
I'm still working on getting the original cassette. No guarantees.

It's a stereo recording of a mono signal. I want to capture only the
information common to both channels and discard any differences as noise.

I should be able to do that with an M-S converter, discarding the side
channel entirely.

Is there a better way?

You've already had a lot of good advice about picking the best channel
and using that, but there is a tool you could try before doing that.
Adobe Audition and its predecessor CoolEdit have a tool called Centre
Channel Extractor. What it does is exactly what you say above. It
examines the stereo signal, amplifies whatever is identical in both
channels and suppresses whatever doesn't correlate - which would be
most of the noise.

It may turn your -15dB noise into -25 or -30dB, which will make a huge
difference when it comes to digital noise removal.

d

I wonder how it works. When you add two random noises together, the power of
the noise goes up 3 dB. When you add two correlated signals together, the
power goes up 6dB. Therefore, the most improvement in S/N that can be had by
summing is 3dB.

There is much more to it than summing, but I have no idea what they
do. You can do pretty much what you want with the centre signal (if it
isn't in the centre you can locate it easily enough) from acapella -
the centre on its own - to karaoke - the centre totally suppressed.

I realize that intelligent strategies exist, but, like the various
restoration programs, they tend to do damage as well. Because of the tape
wiggle, it's hard to see how the program could have accurate enough
information to work on.


There is a fix for that too. Audition will mend minor phase changes
between channels and line them up for you.

I can see how the product you mention could be useful if the primary
problem was dropout.

Bob Morein
(310) 237-6511





If the primary problem is dropout, that requires a very different
solution. Fortunately a stereo source provides this because the
chances are good that the dropout only hits one track at a time.
Selective panning before reduction to mono will do the job. Very
time-intensive though.

d

[Carey Carlan]

(Scott Dorsey) wrote in :

You really, really want the original.

Agreed. The hard part is finding them, if they even exist anymore.

[Carey Carlan]

(Don Pearce) wrote in news:49477b34.503246718@localhost:

You've already had a lot of good advice about picking the best channel
and using that, but there is a tool you could try before doing that.
Adobe Audition and its predecessor CoolEdit have a tool called Centre
Channel Extractor. What it does is exactly what you say above. It
examines the stereo signal, amplifies whatever is identical in both
channels and suppresses whatever doesn't correlate - which would be
most of the noise.

Thanks, Don. I have Audition and am playing with the Center Channel
Extractor. It doesn't seem to be doing much good so far. I'm guessing it
expects a bit more civilized signal--not so far out of phase and with less
noise. I'll keep fiddling and see what I get.

[Don Pearce[_2_]]

On Wed, 17 Dec 2008 14:05:40 GMT, Carey Carlan
wrote:

(Don Pearce) wrote in news:49477b34.503246718@localhost:

You've already had a lot of good advice about picking the best channel
and using that, but there is a tool you could try before doing that.
Adobe Audition and its predecessor CoolEdit have a tool called Centre
Channel Extractor. What it does is exactly what you say above. It
examines the stereo signal, amplifies whatever is identical in both
channels and suppresses whatever doesn't correlate - which would be
most of the noise.

Thanks, Don. I have Audition and am playing with the Center Channel
Extractor. It doesn't seem to be doing much good so far. I'm guessing it
expects a bit more civilized signal--not so far out of phase and with less
noise. I'll keep fiddling and see what I get.

Just had a try with it myself, and it isn't great with broadband
noise. When you do get it suppressing, it makes the noise go very
swirly and obvious. May be I just didn't play long enough though. I
used the acapella preset.

d

[Carey Carlan]

(Don Pearce) wrote in news:49490808.604835515@localhost:

On Wed, 17 Dec 2008 14:05:40 GMT, Carey Carlan
wrote:

(Don Pearce) wrote in news:49477b34.503246718@localhost:

You've already had a lot of good advice about picking the best
channel and using that, but there is a tool you could try before
doing that. Adobe Audition and its predecessor CoolEdit have a tool
called Centre Channel Extractor. What it does is exactly what you
say above. It examines the stereo signal, amplifies whatever is
identical in both channels and suppresses whatever doesn't correlate
- which would be most of the noise.

Thanks, Don. I have Audition and am playing with the Center Channel
Extractor. It doesn't seem to be doing much good so far. I'm
guessing it expects a bit more civilized signal--not so far out of
phase and with less noise. I'll keep fiddling and see what I get.

Just had a try with it myself, and it isn't great with broadband
noise. When you do get it suppressing, it makes the noise go very
swirly and obvious. May be I just didn't play long enough though. I
used the acapella preset.

That appears to be a pretty severe setting.

The biggest problem I find is that it has to be the last processing
step. Even at moderate settings it leaves the phasing so messed up that
any further processing sinks everything into Charybdis.