Bernie,
Here's my thinking on this problem. If you had captured a stereo recording
instead of the mono recording that you got, you would have captured
directional information as well as ambient information. The directional
information is lost. If there were two sound sources, one of them to the
left of the other, only you will be able to tell that! If we give up on
that lost directional information, there are still possibilities for making
the recording sound more real than it does in mono, and that would be by
using convolutional reverberation. Of course the recording already contain=
s
reverberation.
What I am suggesting is that you find a local environment which is similar
to the environment in which you made the original recording, and that you
then acquire stereo impulse responses in that environment. Apply whatever
is the most promising set of stereo impulse responses to your mono
recording, and you will have added a stereo perspective. Of course, it wil=
l
only be a stereo perspective on a single channel source.
There are hope for improved techniques in the near future. Research into
blind source separation is promising, although I'm not aware of anything
that is "ready for prime time". I have heard good demonstrations of
disassembling of music into its component parts, parts in the musical sense=
.
But I do believe that in the not too distant future it should be possible t=
o
separate a mono recording into its perceptually relevant parts. Such as 1)
first animal vocalization 2) second animal vocalization 3) drone of insect
one 4) drone of insect two, ambient sound (stream rushing) and so on. At
that point, you will be able to dissect your recording, and then to
reassemble in a fashion that more closely sounds like what you hear, when
you were there.
If you are interested in following up on the convolutional techniques, but
if you haven't done something like that before, feel free to contact me off
list.
Eric
|