Re: [hatari-devel] FW: Stereo YM2149

[ Thread Index | Date Index | More lists.tuxfamily.org/hatari-devel Archives ]


----- Nicolas Pomarède wrote:
> Le 25/09/2015 16:08, Konador, Cyprian a écrit :
> > Hi All,
> >
> > There is a cool stereo modification for ST:
> > http://rdemming.home.xs4all.nl/Atari/StereoST/
> > I would say that YM in stereo sound really cool.
> > Under Windws/Linux you can check that with Ay_Emul
> > http://bulba.untergrund.net/emulator_e.htm
> > It can play in stereo chiptunes in YM2, YM5 and YM6 format.
> >
> > What do you think about adding that YM stereo into Hatari?
> 
> Hi
> 
> I know this modification, a similar one was in French ST Magazine
> more than 20 years ago, and I modified my STF to support it :)
> 
> As for adding it to Hatari, I don't really know ; the problem is that 
> musics were not composed with stereo in mind, so in some cases
> one instrument will play on voice A, then voice B , then ...
> 
> So if you decide to mix A+B on left and B+C on right, results are likely 
> to vary a lot. Some songs will sound good, some will sound weird. It's 
> nearly impossible to find a unique combination for all cases, so that 
> would need quite some changes in the UI to do your own L/R mixing.
> 
> But then, a much more annoying problem is that sample sound will not 
> sound correct after that, because replays expect to mix the 3 voices 
> into one mono output to get the best possible result with only 16 levels 
> of volume per voice. If you split into L/R, then you will get 2 parts of 
> the same sample on each L/R channel, but with lowered quality.
> 
> All in all, from my own STF experience with this modification, although 
> it looked cool at first, some musics will sound nice, but some will 
> sound worse and in the end I'm not sure it's worth the trouble for 
> Hatari (except is someone is willing to send a patch for this :) )
> 

It will be possible to mix two YM2149 channels in different
combinations without making a new sample table if one
makes a two channel Math model (3 channel model exists).

Another method is to double the channels of Hatari's
sound by copying and doubling the existing sound code.
Then one could mix the channels at the very end. This
would give many possibilities (2 x YM2149 + 2 x LMC1992)

One thing that I observed but never mentioned, is that the
ST uses a 0.33uF Ym2149 capacitor and sounds slightly
muffled. The STE uses 0.10uF and sounds slightly dry.

Maybe a 0.22uF capacitor will sound perfectly Wet.

Note that I am not in favor of turning Hatari into a monstrosity.
I would rather Hatari were a collection of emulators than
an over-sized Swiss army knife. Code maintenance is not
a problem because one should immediately face the
consequences. Hatari is already two emulators, this is good.


> It's sure we can add a lot a filter / effects for sound that were not in 
> the original HW, but this requires time and in my case I prefer 
> improving accuracy for now :)
> 
> Nicolas
> 
> 



Mail converted by MHonArc 2.6.19+ http://listengine.tuxfamily.org/