FWIW I pretty much agree with what Tom Fine says. In addition I sent these
remarks over to old friend Tony Lauck, who independently has been
researching this topic. He returned with:

EQ can be fixed in the digital domain if it was done incorrectly, but
results won't be as good as if correct equalization had been used prior to
Ticks and pops must be removed by hand. It can be an artistic judgment
whether they belong in or out, e.g. bows striking music stands. I use
iZotope RX 2. On really bad disks automatic click reduction can be used on
a conservative setting to get rid of the worst half of the noises, leaving
less manual work.  (On the worst recordings it can take an hour's work to
clean up a minute of music.)  This software can also fake tape dropouts,
sometimes up to 1/4 second. It can even fake glissando and vibrato that are
missing, but this doesn't always work correctly.

Digital noise reduction can be used to remove constant noise, which
includes tape hiss, constant room machinery and hum. Good noise reduction
software (iZotope RX 2 mode C) doesn't degrade the musical sound despite
stripping the noise, but there will be a noticeable loss of ambiance if
much noise reduction is used. Whether this is advantageous or not will very
much depend on the particular recording. Unfortunately, this style of
psycho-acoustic noise reduction doesn't work with periodic noises, which
include ignition noise or 78 RPM swoosh, because the algorithm use a noise
"profile" to define what is noise and what is signal. If the profile is set
on the quiet portion of the swoosh it will leave the loud noise in. If the
profile is set on the noisy portion of the swoosh it will scrape away some
of the music. (Perhaps some software can be developed to deal with this
situation, but I've not seen it.)

It might seem that one could tackle a recording one second at a time, but
it turns out this strategy doesn't always work. If there are many problems
in a transfer that get "fixed" the net result can be weird even though each
individual change seems like a good one. It may be that listening to 10
seconds sounds good, but if one listens for minutes one gets a strange
feeling that something is very wrong. This is particularly true when
removing acoustic noises. (Example: a lecture where there were many people
coughing. It was possible to remove coughs between the speaker's syllables,
but the coughs that overlapped were left in. Any short period sounded
natural, but after listening for more than about 30 seconds it became
obvious that the audience only coughed when the speaker was vocalizing.


TOTALLY agree with that observation.


On Tue, Apr 10, 2012 at 8:49 AM, Tom Fine <[log in to unmask]>wrote:

> Read the article, it was in the ARSC Journal a while back, maybe a year or
> two. Too technical to summarize here, at least for this non-EE.
> ----- Original Message ----- From: "Andrew Hamilton" <[log in to unmask]>
> To: <[log in to unmask]>
> Sent: Tuesday, April 10, 2012 8:39 AM
> Subject: Re: [ARSCLIST] Recording_78rpm_records
>  Gary has also written for ARSC Journal advocating analog playback and EQ
>>>>> of
>>>>> grooved media, rather
>>>>> than "flat playback" and software EQ, and has specified the technical
>>>>> reasons why analog EQ works
>>>>> differently from DSP EQ.
>> What is it about minimum phase digital audio equalization, such as that
>> found in a de-RIAA or RIAA plugin, that Mr. Galo feels is not up to snuff?
>> Andrew