[hatari-devel] Choosing ym mixing method in the UI

[Anders Eriksson]

On Mon, 4 Apr 2011, Eero Tamminen wrote:

>> ------------------------------------------
>>
>> YM voice mixing  [] linear   [] ST table
>>
>> ------------------------------------------
>
> Isn't the table used for mixing STE/Falcon/TT YM sound?

Hello,

no this is the mixing of the three YM channels.

On the Atari computers they are mixed simply by joining the outputs, which 
brings some problems. Say if channel A plays at volume 15, it will 
feedback into channel B and C to produce further sound. So the mixing of 
the YM channels can't be as simple as A+B+C if you want accurate results.

To play samples on the YM, you combine volumes of all three channels (for 
the better quality players) and with the cheap mixing of the ST, the table 
of volume variations is nothing like linear. We did some simple analysis 
of three volume tables to see how they really look, here's the result:

http://files.dhs.nu/files_coding/volumes.pdf

The blue volume table is the one used in Protracker ST, the yellow one 
from Quartet and the red one which is linear (works for other computers 
with proper channel mixing, such as the MSX, CPC, Spectrum..).

The Protracker table sounds by far the best on a real ST, and the curve is 
really obscure :)  It's this behaviour that Hatari needs to simulate to 
accurately replay YM-samples. The red linear table is HORROR on a real ST 
and perfect in Hatari with linear mixing :-)

If you want to hear the Protracker volume table on Hatari vs the real 
thing, check out our latest intro:

http://files.dhs.nu/demos/sh11inv.zip (executable)
http://dhs.nu/video.php?ID=317 (video from a real STe)

I think even you would agree the linear Hatari mixing needs fixing :-)


--
Anders Eriksson
ae at dhs.nu     http://www.dhs.nu/
ae at atari.org  http://www.atari.org/