Previous Thread
Next Thread
Print Thread
#71144 06/24/11 08:18 PM
Joined: Feb 2006
Posts: 30
Member
OP Offline
Member
Joined: Feb 2006
Posts: 30
Taken from the SPU in my SCPH-5501, read internally digitally.

General channel playback, 256 phases, 4 "taps" = 1024 coefficients: http://pastebin.com/raw.php?i=ACCKWy0y

44100 -> 22050 dowsampling for reverb(might be the same for upsampling, haven't tested, but the zeroes make for nice optimization possibilities):

Code:
 static const int16 ResampTable[40] =
 {
  0xffff,
  0x0000,
  0x0002,
  0x0000,
  0xfff6,
  0x0000,
  0x0023,
  0x0000,
  0xff99,
  0x0000,
  0x010a,
  0x0000,
  0xfd98,
  0x0000,
  0x0534,
  0x0000,
  0xf470,
  0x0000,
  0x2806,
  0x4000,
  0x2806,
  0x0000,
  0xf470,
  0x0000,
  0x0534,
  0x0000,
  0xfd98,
  0x0000,
  0x010a,
  0x0000,
  0xff99,
  0x0000,
  0x0023,
  0x0000,
  0xfff6,
  0x0000,
  0x0002,
  0x0000,
  0xffff,				     
  0x0000,
 };

(Really only 39, but 40 here for consistency with my code and for potential future SIMD usage).

>> 15 to remove fractional components for the output after doing MAC, for both tables.


I can't guarantee they're 100% correct, though they should be unless I missed or misinterpreted something. Ideally someone else would do the same, and we could compare and investigate any discrepancies. Using them in combination with the proper algorithms is certainly more accurate than something naive like linear interpolation, or using the SNES sound chip coefficients(unless there was a revision of the SPU in which they were the same, which I suppose is possible...).

Last edited by Mednafen; 06/24/11 08:20 PM.
Joined: Mar 2001
Posts: 16,613
R
Very Senior Member
Offline
Very Senior Member
R
Joined: Mar 2001
Posts: 16,613
Nice work. I'm making sure pSXAuthor sees this so he can update his SPU core (which is also what we use in MAME now).

Joined: Aug 2008
Posts: 14
N
Member
Offline
Member
N
Joined: Aug 2008
Posts: 14
if only neill corlett wasn't so stingy abt highly experimental...

Joined: Mar 2001
Posts: 16,613
R
Very Senior Member
Offline
Very Senior Member
R
Joined: Mar 2001
Posts: 16,613
HE uses the SNES coefficients - Corlett was the *source* of people doing that.

Joined: Aug 2008
Posts: 14
N
Member
Offline
Member
N
Joined: Aug 2008
Posts: 14
double aaugh. i'm still not holding my breath for the day he finally open-sources it so i could study it; aosdk is a little too unwieldy for me at the moment.

Joined: Mar 2001
Posts: 16,613
R
Very Senior Member
Offline
Very Senior Member
R
Joined: Mar 2001
Posts: 16,613
I'm not sure he can - I'm pretty sure Highly Experimental contains Sony ROM images. The "unwieldy" part of AOSDK (and SexyPSF, which is structured similarly) is the code to HLE the Sony OS functions. Both players would be very little code using real ROMs smile

(I'll also note it's not *that* unwieldy given there are successful ports of it on Android and iOS).


Link Copied to Clipboard
Who's Online Now
1 members (r09), 36 guests, and 3 robots.
Key: Admin, Global Mod, Mod
ShoutChat
Comment Guidelines: Do post respectful and insightful comments. Don't flame, hate, spam.
Forum Statistics
Forums9
Topics8,836
Posts116,222
Members4,921
Most Online890
Jan 17th, 2020
Powered by UBB.threads™ PHP Forum Software 7.7.5