I wrote:

> Not quite enough to decode.

Actually it did decode with a sneaky -f9 option
to the Linux decoder.

Rank 440334 Eb/N0 = -0.2 BER 40.0% phase 0 0 30 30

and

Rank 218308 Eb/N0 = -0.9 BER 40.8% phase 30 30 60 60

so a slight frequency offset perhaps?

The -f9 option tells the decoder to output all decodes
found in the list.  By default it stops at the first
(which in this case is a false decode).

The correct decode is actually the 2nd strongest
in Eb/N0 ranking, 3rd strongest by path metric,
after a full phase search.

I didn't notice any timing offset when I examined the
previous WAV file from Alex (Feb 28th 8chmessage.wav).


--
Paul Nicholson
--