I don't think impedance matching is as much of an issue as the level.
From
what I can tell, Bill had it right. All it takes is a couple of
resistors.
You got to reduce the level A BUNCH. I don't know if Bill's 100:1 is
right,
I found the values for my Icom by experimentation, but it's pretty close.

Keep in mind that you want to run the computer's speaker volume pretty
high
to reduce quantization error. If you adjust the resistance and mike gain
so
that your power out is somewhat below max when you are generating SSTV, it
should be pretty close.

Digipan suggests a 100K in series and a 1K in parallel ... seems like I
had
to get more extreme, but then I'm connecting up an Icom.
========================
Apart from (roughly ) matching impedances ,a transformer ensures galvanic
separation between PC (laptop) and transceiver , hence avoiding any hassle
due to earth loops .
If the voltage level at the transceiver's microphone input poses a problem ,
a resistive voltage divider can always be added.

Frank GM0CSZ / KN6WH