I looked briefly at the Stanford piano roll processing software.
It is reasonably commented but it would take many hours to fully
understand it.

I think that the basic image processing is probably correct but it's
a mystery to me why they felt they needed to write their own when
there is already excellent processing software by Warren Trachtman and
also by Anthony Robinson and Wayne Stahnke.  Must be the "not invented
here" university syndrome.

My first glance at the emulation part of the code suggested that it is
woefully inadequate.  Again, there is excellent emulation software for
Welte and Ampico written by Wayne Stahnke.  I believe he even offered
his software to Stanford and was rejected.

So I decided to listen to a few pieces, just on the computer playing
with Pianoteq to provide a reasonable piano-like sound.  I can possibly
find time to play some more on my concert grand piano and also possibly
compare emulations of the same piece made by Stanford and by Stahnke.

The files are named with the roll number at the end of the name, which
is pretty stupid and makes it hard to find specific files.  You have to
sort by file name and use the (meaningless to anyone but the people who
made the names) prefix to narrow down the search.  I'm mentioning this
just as an example at being not user-friendly.

So far I have listened to part of one file only, the Grieg Piano
Concerto in A Minor.  I am not familiar with the Welte version but
I know the Duo-Art quite well as I listened to it many times when
I had a Duo-Art piano.

In the Duo-Art piano, the dynamics are stunning and chilling.  Playing
the Stanford Welte version I get the impression -- as I do with most
attempts at emulating (including most of mine done with early software)
-- that there is no reason to listen to this crap.  It is flat and when
it's not flat it's rather random in the dynamics.  I do want to restate
that I just listened to a small part of the file on a computer and not
a piano but I doubt that my piano is going to magically make it sound
good.

I will listen to more if I can stand doing so but my initial impression
is that the emulation aspect of the Stanford software is a simplistic
amateur attempt and is essentially worthless.  I am suggesting that
anyone who wants good emulations might give a second thought to trying
to compile and figure out this software.

There are fine alternatives that work really well: Wayne Stahnke's
emulation software for Welte and Ampico and Richard Brandle's Wind for
Duo-Art.  I invite others to listen to the Stanford emulations and give
their opinion.

Spencer Chase
Garberville, California