Re: OT: singing training

[Graham Percival]

On Mon, 15 Sep 2008 22:06:00 +0100
Nick Bailey <address@hidden> wrote:

> I'd be a bit careful about teaching absolute pitch accuracy. I'm not  
> a musician, but it is a common mistake by us engineers and computer  
> scientists to consider individual notes important. 

Relative pitch accuracy can still be done -- after the note
segmentation, take the average pitch[1] of the first note, then
compare all other note pitches to the first one.  The student
could then perform the exercise at whatever pitch was comfortable.

[1] however we define "average" in this case -- whether we ignore
outliers, take a weighted average over time, etc.

That said, (classical) singers *do* need accurate pitch accuracy
when singing with a piano.  Of course, the piano helps the
singer's pitch a lot.  To mimic that on a computer, I'd propose
something like a midi piano (listened via headphones) to accompany
the student's singing (to be picked up via a microphone).

...
Now, this still doesn't specify what our hypothetical program
should do when it detects a note badly out of tune.  If the
student sings a single note out of tune, we don't necessarily need
to warn them (or give them a bad grade).  As a cello teacher, I'd
still want the program to warn students (and perhaps scale the
grade penalty based on the duration of the note), but a singing
teacher may have different requirements.

> You might find they prefer your daughter to work from a primer
> which develops pitch accuracy  alongside other musical skills.
> Again, IANAMusician, but I suspect  you are doing a lot of good
> in getting your daughter to interact with other players (you)
> instead of a mechanical device, so perhaps it's a case of "be
> careful what you wish for, you might get it".

This is one advantage of highly targeted exercises (like MEAWS),
as opposed to some other Computer-Assisted Instruction projects
which attempt to do everything at once -- if your exercises
only consist of 8 quarter notes, then nobody (be it teachers,
parents, or students) could possibly mistake the computer program
as covering all aspects of music.

That fact that this enormously simplifies the audio analysis as
well is simply a nice bonus.  My main concern, like you, is that
some people might rely on the computer program to cover all
aspects of music instruction, and we're at least 50 years away
from *that* happened.


> You should also not underestimate the importance of rhythmical
> accuracy.  Perhaps it is better to impose a tempo, even a slow
> one, from scratch. Give a pro and a less experienced musician a
> piece to play, and I bet the difference will manifest itself
> more in rhythmic accuracy than in intonation.

Definitely!

By the way, Tim, the rhythmic portion of MEAWS will work just fine
with singers.  We all sound the same when we clap!  :)
http://web.uvic.ca/~gperciva/meaws/


Cheers,
- Graham