Re: [hatari-devel] Gain too low in DmaSnd_LowPassFilter for STE ?

[David Savinkoff <dsavnkff@xxxxxxxxx>]

Hi,

Well... at least I got dmaSnd.c right.
Now I'll take my time to get more done right.


----- Nicolas Pomarède wrote:
> Le 07/08/2016 à 20:45, David Savinkoff a écrit :
> >
> > Do this instead.
> >  /****** return (in * 3) << 1;  */
> > and not what I said earlier,
> > but experiment
> > gotta go.
> >
> 
> 
> hi
> 
> "return (in * 3) << 1"  (which is "in *6" in fact :) ) gives much more 
> hearable digi sound in the case of these musics, this sound better.

Louder is indeed better! :)

> 
> As for really fine tuning the YM and DMA sound relative output volume on 
> STE, I think we would need to record some STE output as a wav file when 
> mixing a simple YM note with a square wave sample for example, then 
> compare the resulting wav with the output of Hatari.
> 
> I will be away for 3 weeks, but if someone feels like recording a real 
> STE, then it could be useful to improve Hatari's relative volume level.
> 
> Nicolas
> 
> 
> 
> > ----- David Savinkoff wrote:
> >> ----- Nicolas Pomarède wrote:
> >>> Hi
> >>>
> >>> as discussed here
> >>> http://atari-forum.com/viewtopic.php?f=16&t=29615&p=299270#p299270 the
> >>> STE DMA sound samples can't be heard when using an output freq < 50066
> >>> Hz in Hatari (for example, try "salamander : planet latis")
> >>>
> >>> At 50066 Hz you hear the samples, but at 48 kHz you don't hear them
> >>> anymore. This is because 50066 explicitely turn DMA low pass filter OFF
> >>> in the code path.
> >>>
> >>> David, do you think it would be possible to increase the gain in
> >>> DmaSnd_LowPassFilter to hear the samples even when using < 50 kHz output
> >>> freq ?
> >>>
> >>> Note that in this music demo, samples already have a low volume (if you
> >>> just listen to the wav file with Audacity for example), so maybe the
> >>> fact that the sample don't use the full 8 bit range is fooling the
> >>> filter and you get an even lower volume in the end ?
> >>>
> >>> Nicolas
> >>>
> >> Hi,
> >>
> >> I will look into this later today. Note that 55 KHz has no filter so
> >> that one can re-sample it with their own filter, thus, 55 KHz
> >> samples should be sharper than a real STE.
> >>
> >> The low pass filter Hatari uses filters high frequencies a little
> >> too much.
> >>
> >> see /****** return (in * 3) << 3;  */ below. Which refers to
> >> hatari/src/dmaSnd.c
> >>
> >> This will increase the gain by approx 3dB
> >>
> >> Experiment until you like it. Note that low frequencies will
> >> be boosted also. Note that just increasing the cutoff
> >> frequency will require an expensive filter.
> >>
> >> /**
> >>  * LowPass Filter Left
> >>  */
> >> static Sint16 DmaSnd_LowPassFilterLeft(Sint16 in)
> >> {
> >> 	static	Sint16	lowPassFilter[2] = { 0, 0 };
> >> 	static	Sint16	out = 0;
> >>
> >> 	if (DmaSnd_LowPass)
> >> 	{
> >> 		out = lowPassFilter[0] + (lowPassFilter[1]<<1) + in;
> >> 		lowPassFilter[0] = lowPassFilter[1];
> >> 		lowPassFilter[1] = in;
> >>
> >> 		return out; /* Filter Gain = 4 */ /****** return (in * 3) << 1;  */
> >> 	}else
> >> 	{
> >> 		return in << 2;  /****** return (in * 3) << 3;  */
> >> 	}
> >> }
> >>
> >> /**
> >>  * LowPass Filter Right
> >>  */
> >> static Sint16 DmaSnd_LowPassFilterRight(Sint16 in)
> >> {
> >> 	static	Sint16	lowPassFilter[2] = { 0, 0 };
> >> 	static	Sint16	out = 0;
> >>
> >> 	if (DmaSnd_LowPass)
> >> 	{
> >> 		out = lowPassFilter[0] + (lowPassFilter[1]<<1) + in;
> >> 		lowPassFilter[0] = lowPassFilter[1];
> >> 		lowPassFilter[1] = in;
> >>
> >> 		return out; /* Filter Gain = 4 */ /****** return (in * 3) << 1;  */
> >> 	}else
> >> 	{
> >> 		return in << 2;  /****** return (in * 3) << 3;  */
> >> 	}
> >> }
> >>
> >>
> >>
> >
> >
> >
> 
> 
>