The reason I used the transformer secondary for an inductor is that this
design requires a huge inductance - around 100mH, rather than 100uH.
That is so I can keep the frequency low, which is what makes the design
so easy to implement.  You could push the frequency of this design to 50
or maybe 100 kHz, with proportional decrease in the inductance, but the
higher you make the frequency, the more attention you have to pay to the
layout, transistor switching speeds, diode reverse recovery time...

It is quite possible that the 2000 Hz ripple will get into the audio.
You'd have to know what it's used for, or just try it and see.  You could
set the switcher for 14V and use a 7912 to remove the ripple.  You could
increase the frequency.  You could improve the output caps.  But maybe
you don't need to.

As for the gory details, I'll have to apologize.  I forgot who I was
talking to.  Since the schematic is too detailed for ASCII art, I'll have
to try to figure out how to get it into electronic form.  If someone
would be willing to receive a fax, scan it and post it, that would be a
lot easier for me.  I'm still in the pre-scanner dark ages.  If you'd
like to volunteer, e-mail me off-line since I don't get the digested
piclist until after midnight!

Don