SpritesMind.Net

TmEE co.(TM) wrote:

Stef wrote:5 Khz is not something which can called "pretty good", but i've to admit that sound not that bad at least

I don't know if you have tested Duke3D on the real thing, but I can say, that it looks and sounds awful on emulators, sound is bad especially on Gens, because its DAC is not very good. On the real thing, its a different story (mostly).

Never tried on the real thing, but tested on Kega.
I'll try it asap.

You'll be impressed how different it looks on emulator and TV. Kega has very good DAC... but it seems to play my WAVPLAYs little faster than on the real thing.

I re-tried on Kega, it's true the sample of Duke Nuken sounds good. I'm really impressed for 5 Khz sample !

See, real system always sounds better than emulation, no matter if you're using the best or worst of them.

I have to disagree, the genesis DAC sounds a bit woofed. Yesterday i did some test of my Z80 wav player (16 Khz sample) and the sound of the genesis was not as good than on emulator imo. For good quality sample, the genesis generally sound not as good than emulator (PC sound card has better output) but for low sample rate the genesis imperfect DAC produce better result and hide efficiently the low rate used

Not true for my system !!! I've modified the audio mixing part and now it has really great sound output, with almost 100% stereo separation. Original sound output system has poor stereo (50% separation) and a bit distorted (very very slightly).

ah ah ah!!
I tested without optimisation of my current code : it support 6KHz sample!!
far from the 16KHz possibility
Ok, I'm playing FM at the same time...perhaps it's the problem...
What if I add PSG support now ?!!!

my my my...Z80 is even harder to optimize than 68k

Well, i just did my first Z80 asm programming these last days and i've to admit i don't like it much... Z80 instruction set isn't the best one :-/
no flag modification for 16 bits arithmetic, 8 bits arithmetic only on register A, no test or tst instruction...
My first program was a 4 bits PCM player on the PSG, but quality was so hurting that i decided to drop back on the DAC. Then my player was able to handle 16 Khz 8 bits sample with some free cpu cycles aside.
Yesterday i tried to play 2 samples at same time with software mixing @16 Khz but for now, my (awful) code can't maintain it and i've some minors bugs to fix. But i think with some optimisations it's something possible

16 Khz gives you 223 cycles for each DAC write

TmEE co.(TM) wrote:Not true for my system !!! I've modified the audio mixing part and now it has really great sound output, with almost 100% stereo separation. Original sound output system has poor stereo (50% separation) and a bit distorted (very very slightly).

You modified the audio output :p
without modification, the sound output isn't that good

Yes I did !!! That hassling with smd components and other tiny stuff was definately worth it. BUT I'm not very food on analog electronics so there are some things which I should redo (improve audio amplifications stuff).
Last days, I've messed with getting decent RGB out of the system, I'd tell more but that's off topic.

ok, i modded KMod
I was able to spy the DAC freq update....

so WAVPLAY is at 11737Hz not 12
and my test is at 6654Hz

KanedaFr wrote:ok, i modded KMod
I was able to spy the DAC freq update....

hehe, neat feature
Anyway playing something > 16 Khz is just a waste :p
I would love to have a z80 code capable of mixing 2 channel at 16 Khz. For now i've 2 channels at 10 Khz...

hop, i won 400Hz...no 7Kz

need to work on the FXaddr++

FXaddr++ ? are you coding the Z80 in C ? that explain some stuff
Maybe your Z80 C compiler isn't a good optimiser...

Just something about the YM2612 : have you to stop the Z80 to access it from the 68000 cpu ? from what i saw, we can access it from A04000 only so we need the Z80 bus... too bad :-/

Why do you need to access YM2612 with 68K and z80 at the same time ?