[LAU] Progressive Quantisation

[Fons Adriaensen]

On Wed, Aug 29, 2007 at 11:19:55PM +0100, Folderol wrote:

> I did some work on PLLs at RF mmmfty mmmf years ago and seem to remember
> that they are only really effective if there is a large difference
> between the controlled signal and the rate of change, which I wouldn't
> have thought would be true in this case.

It's not clear what exactly you mean, but anyway a the working of 
a PLL does not depend on 'large differences' - it can track even
the most subtle changes.

> This actually, isn't really what I'm after anyway. I want to do what is
> potentially much simpler, I want to manage a completed MIDI recording
> rather than working on-the-fly.

For a software PLL it doesn't matter if it operates real-time or on 
stored data. But the latter makes it possible to use more advanced
systems, using things like  open-loop algorithms (which can very suble
changes and work in non-linear ways without the risk of instability),
or backtracking - reviewing previous decisions in the light of new
information.

But whatever approach you use, if it is capable of tracking tempo
changes it will be a form of PLL, but maybe in disguise. 

-- 
FA

Follie! Follie! Delirio vano è questo !