Grity:
if you need stady cruise for your intresting experiments, did you think it can be possible to create a system which can read the birds noise and extract the speed (or altitude)(or even the change of this parameters) while it compared the currend sound profil with a databank of sounds ?
What you need to know is at least one significant data point (be it altitude OR speed) and have a set of profiles over a reasonable range (something like 20kt speed increments, or 1000ft altitude steps), and perform a best match, which in my case would be least residual... the profile subtraction that resolved to the least residual would give you ballpark parameters.
The reason you need to know at least one key parameter is due to the fact that cruise at FL350 M0.79 can easily look the same as cruise at FL380 at M0.76 for example, at least as far as noise profile is concerned. If you had a LOT of data samples to work with you could refine this, but generally I'm lucky to have 3 or 4 samples of cruise at various specific alt/speed combinations, not the 10's of samples that would really be needed.
As for a next generation pinger design, it would make a lot of sense if it were a transponder design - from what I can find sprinkled around the web, this should result in approximately double or better the range being possible.