[sdlsound] alt audio converter

[Frank Ranostaj]

> though the result certainly sounds... interesting.
> 
> Please note that I didn't make an exhaustive test of every possible code
> path. What we really need here is a complete set of test cases, and a script
> to automate them. But that's for another day.

Ok. I've found - and fixed - two serious bugs. Now the rate converters
produce
something sensible. The framework seems to be in place now.
On the other hand I am not to happy with the design direction, it might
be
more useful to do an accurate sample rate conversion and not one with
0.3%
frequency deviation. An accurate conversion leads to an approx. 4 time
higher 
calculation load. May I ask for comments? 

> Despite these complaints, let me emphasise that I am pleased and excited
> about the prospect of new and improved conversion filters. They just need a
> few more bugfixes, that's all. :-)

There are probably a handful (or two) of by one errors left. I am not
really
sure if I got the variable rate converter right, i.e. if the phase steps
are in sync with the selection and movement of the filter.
A good test signal would be a sinus of 0.9 times the max. dest.
frequency, 
and another test signal could be a sinus of 1.01 times max. dest.
frequency.
E.g. if we convert from fs = 22050 Hz down to fs = 8000 Hz, we want to
try a 
signal of 3600 Hz and one with 4040 Hz. 
There might be some (these are easy to fix) scale issues, too.
Finally I am not to happy with the 'for' loops in the
alt_audio_converter.c
nearly each one has a different structure.

Anyway, I've to do some other work for next couple of weeks, so this is
my 
last update for now. After rejoining I'll add some documentation to
remove 
some of the 'black magic' :-)

	Frank
-------------- next part --------------
A non-text attachment was scrubbed...
Name: audio.zip
Type: application/x-zip-compressed
Size: 10629 bytes
Desc: not available
URL: <http://icculus.org/pipermail/sdlsound/attachments/20020622/654fbeb5/attachment.bin>