Is the PSG worth it?
Edit:
OK OK, I’m convinced :) PSG support will come! (sooner or later)
The PSG (programmable sound generator) is the thing that produces the Game Boy’s typical sound, which is still present in the DS (or so it seems). I have promised to include support for it in NitroTracker several times now. But after having seen the Game Boy Sound Comparison, I suspect it might be emulated in software. If that is the case, there wouldn’t be a point in going through the pain of implementing PSG support, but it would be much easier to generate the sounds in software.
Long story short: I’m looking for people with good ears who think they can tell real PSG sound from a cheap emulation :-) I wrote a test program that generates the same tone sequence two times: One time using the PSG and the other time using software. I’m not telling you the order though :-) If you have a DS, please download it here and tell me in the comments whether or not you hear a difference, and if you do, which one you think is produced by the PSG.
If you don’t own a DS, you can listen to the samples below instead. I intentionally didn’t label the sets as PSG or software, so you have to find out for yourself. Please post your guess in the comments, and whether or not you could hear a difference.
To preserve the sound, I didn’t compress the samples, so they are rather big (1.4MB each).
Set A
Pulse 1
Pulse 2
Pulse 3
Pulse 4
Noise
Set B
Pulse 1
Pulse 2
Pulse 3
Pulse 4
Noise
Here are some closeups of the puse waves:
A
B
Thanks!
Tob
April 25th, 2007 at 8:18 am
They’re very close. I think real PSG is B, as is seems there’s a little phase effect and some thin distorsion, and A versions sounds much clearer, there’s less “analog feel” in my sense. But still, it’s very close.
May 14th, 2007 at 10:26 pm
Well, I can’t tell the difference. If it were me, I would just implement whichever was the easiest.
BTW, thanks a lot for making nitrotracker.
It’s the best homebrew app out.
May 21st, 2007 at 5:29 am
Listening to the WAV files through my computer I can’t really tell the difference, but through my DS Lite the second sample sounds a lot cleaner, the first having a little bit of background noise on the higher tones. I would assume emulation would produce a cleaner sound, so I’m going to have to go with the first one being PSG. I don’t mind which way you go with Nitrotracker, but the sooner the better ;) (I’m really looking forward to it).
Keep up the great work.
June 26th, 2007 at 1:24 am
Tested on DS lite, with an M3lite…
0xtob! Your PCM fakery of the PSG channels is perfect (or at least my eardrums and auditory circits tell me) for:
[Pulse 2] 25%
[Pulse 3] 37.5% and
[Pulse 4] 50%
but Set B 12.5% pulse [Pulse 1] is a bit fuzzier than Set A’s.
isn’t set B the PCM fake? =P
Noise sounds off…. but I understand it’s hard… noise channels somehow will always sound best in hardware…
July 8th, 2007 at 9:12 am
Hey, thank you all for listening!
To solve the riddle:
In the test app, the first set of sounds is the PSG and the second set is fake.
In this page, it’s just the other way around :)
For faking the PSG, I used the info from here:
http://nocash.emubase.de/gbatek.htm#dssoundnotes
As I said, I will go with the real PSG, but it won’t be in the next version unfortunately.
Thank you all for your nice comments!
October 27th, 2007 at 12:46 pm
Hey, I think PSG is a good choice, but if you were to do it in the dac with your own emulation, you could put other emulations in there, too: like sid, atari, nes, etc. Just a thought. If they aren’t sample-based, it wouldn’t be too hard (much of the code is probably open source).