Thanks for the replies again
Funnily enough I can find nothing in the literature dealing with how the
curve should be drawn, pretty much every text assumes the line will be drawn
by hand to a "best fit curve". I'll give a go to the linest and trend
functions over the next few days and report back. I must say though I was a
little dismayed by the formula for calculating polynomial regression, I
rather thought before I got started on this that working out how to get excel
to computerise the analysis of speech audios would be pretty straightforward
and take me an hour or two, instead I've opend the lid of some terrible
pandoras box of statistics!