Re: (OT) MEAWS user study

[Graham Percival]

On Wed, 23 Jul 2008 10:51:42 -0700
Patrick Horgan <address@hidden> wrote:

> Has anyone tried the violin part as a vocal sight singing training
> tool?

Not that I know of, although you're the third person to mention it
(unoffically).  I hope that you'll fill out the feedback form as
well, so I can officially discuss it in the thesis.  :)

I see three problems:
- the range.  MEAWS violin intontation is aimed at violin, mostly
  on the upper two strings.  The first exercise goes from A440 to
the E above it.  That might be ok for sopranos, but that's it.

- the intervals.  The first intonation exercise is nothing but
  leaps of a fifth.  For violins, this is trivial: it's just
changing between open strings.  For singers, this would probably
be exhausting, and is certainly not the easiest thing to begin
with!

- the audio analysis.  MEAWS uses an extremely simple algorithm
  for grouping pitch-frames into notes (the reasons for this
simplicity are given in chapter 1).  This algorithm works great
for violins, where there's a sharp gap in the pitches.  For
example, when you play a different string on a violin, there's a
very sharp boundary.  The human voice doesn't do the same kind of
thing.


Given the apparent interest in sight-singing practice, this may
well be something to include in my PhD, if I can continue working
in the same area (all depends on the
university/supervisor/funding, all of which will be investigated
and hopefully determined in the fall).

Cheers,
- Graham