Reply 100 of 148, by Snover
- Rank
- l33t++
I see you've started emulating the reverb but there is no chorus component on the list of components. What gives?
Yes, it’s my fault.
I see you've started emulating the reverb but there is no chorus component on the list of components. What gives?
Yes, it’s my fault.
For all the documentation I have on the MT-32, there is no mention of any chorus or chorus control. Its my assumption actually that the MT-32 doesn't have a chorus effect. The D-50 did, but the MT-32 doesn't.
Oh... about keeping the older sample files as "documentation" or even posterity: No, I haven't been. They sounded hideous and I want to keep both my hearing and my speakers intact.
Against my will (this MT-32 thing is addictive!) I will be posting another update tonight. This one will have volumetric reverb and a few corrections on the instruments.
Originally posted by canadacow
I've got the filter correctly implemented and the waveforms all being generated to spec. Though they may see a little more tweaking in the future, this is more or less it. The only things missing now are reverb and chorus, which when complete, should reproduce the authentic MT-32 sound! Whoohoo! Stay tuned!
and
Originally posted by canadacow […]
Originally posted by canadacow
Originally posted by Snover
Why are the first notes so nasty on the emulator?
Again, its the missing chorus. Once that's in place it should sound fine.
hmm, hmm... 🙄
Oh, BTW, 3000th post, w00t 😉
Yes, it’s my fault.
Yup. I was wrong. In writing my emulator its acutally quite insane how many times I've been wrong about something.
Cancel that earlier message about a new version tonight. I've decided instead to hold off on new releases until I get the PCM partial numbers matched up correctly. Its very frustrating... I can listen to the analogue synthesis and hear almost identical MT-32 emulation. On the other hand, when I turn on the PCM partial components, everything goes to crap as the wrong PCM parts are playing and screwing up the instruments. Its kind of embarassing as it ruins what otherwise would be decent emulation (particularlly with my volumetric reverb turned on!). Oh well.. I have another highest bid on an MT-32 on eBay.
Ran the MT-32 version of DOSBox with SQ3, not bad at all, but the stereo mixing isn't there. On the intro the drums quite clearly pan from right to left, but do not in the emulator.
Sounds pretty damn good, though, I must say.
Oh, and I wouldn't want a warrantee on my DOSBox, as nice as I'm sure the person is, thanks. *cough*Warranty*cough*
Yes, it’s my fault.
Thanks for the compliments. It sounds ok. Its still really just a "good start". When I compare my emulator to the file posted by Spikey, I realize just how much further I have to go. I'm starting to get really frustrated though. I'm getting to the point were I need to see specifically what the MT-32 does for certain things so I can ultimately tweak the parameters just right. I keep bidding on MT-32's on eBay and they keep going out of my price range (I am after all, trying to write an emulator to eliminate the necessity of having the physical hardware). I don't need a digital ripping rig. All I need are some sound files that demonstrate how the filter parameters are interpreted. Vladr, any progress on those PCM samples?
Couple of points:
No chorus on the MT-32. Inbuilt if it exists at all.
And my mp3 was ordinary, I think excess reverb.
Anyways, here's something more realistic to aim for:
Leisure Suit Larry 1 VGA- All Tied Up!
Recorded from the Roland MT-32, reverb 5, 44.1 KHz, 128 kbps.
- Alistair
Hmmm... no MIDI file for that song on Quest Studios otherwise I'd post an emulated MP3 for you.
Good news: Got an MT-32 on its way to me in the mail. Should have it by tomorrow. Hah... as luck would have it, tomorrow is also my wife's birthday so it'll really be the weekend before I'm allowed to play with it. 😁
Married in college, eh?
...
okee.
Yes, it’s my fault.
Graduate school.... its completely different than college. Is more like a job than school at this level (I even get a stipend just for showing up to class!!). Here's the real shocker... my area of study...?? Sociology! (Maybe that explains why I don't document!) 😜
Just stay clear of those USDA officials. 😜
Damn canadacow I forgot! Sorry.
Attached to this message. 😀
And a MT-32, nice! It rules.
- Alistair
P.S. Actually I think it's reverb 8, not 5 (the mp3). If it matters. 😀
Well... been playing with my MT-32. Wow, boy do I have a long way to go. Just concerning the PCM partials...uhm... apparently the MT-32 plays them all at different base rates and then it loops a few. I still can't get a few of them to match up. I haven't even begun to mess with the filter. Sheeesh.
wrote:Here's the real shocker... my area of study...?? Sociology! (Maybe that explains why I don't document!) 😜
Actually, one of the rare things you documented in your MT-32 source code is the fact that you are studying Sociology... 😜
Yeah... because they tend not to mess with Digital Signal Processing much when talking about Marx's theories.
Glad to see someone's actually looking at my code though. It'll really make me happy if someone were to perk up and give suggestions beyond "document more." By the way, since I got my MT-32 I have realized several things:
The cutoff freq and filter envelope are independent of each other (well, in the sense that the filter goes on top of the cutoff freq). Most surprising is the envelope delays themselves. Its exponential. 100 = about 6 seconds. 75 = about 1.5 seconds. 50 = .75 seconds... 25 = .375 seconds, etc.. you get the idea. This is going to add a serious level of complexity to my emulation. Whoohoo.
wrote:Yeah... because they tend not to mess with Digital Signal Processing much when talking about Marx's theories.
Glad to see someone's actually looking at my code though. It'll really make me happy if someone were to perk up and give suggestions beyond "document more."
That won't be me, I have only very basic coding notions... I haven't been able to even compile DOSBox with your MT-32 emulator (I can't find mathutils.h)
Most surprising is the envelope delays themselves. Its exponential. 100 = about 6 seconds. 75 = about 1.5 seconds. 50 = .75 seconds... 25 = .375 seconds, etc.. you get the idea. This is going to add a serious level of complexity to my emulation. Whoohoo.
If it was exponential, it should be transformed to something like this :
delay = 6*2^((x-100)/25)
where 0<=x<=100.
[size=0]Edit: I found the constants :p, quite easy in fact[/size]
But what you describe is not really exponential: between 100 and 75, the delay is divided by 4, then between 75 and 50, then 50 and 25, it's divided by 2 each time...
I'm not even sure if I use anything from mathutils.h in my latest source code. Oh well, there's a binary available. My next version (when school is out) will have a quite a few changes to the source code so it'll be a while
Just a quick update for all those who care:
I have the Windows Driver Development Kit on order and when I get it I'll be taking the emulation code out of DosBox and making it into a generic WDM and DirectMusic driver. I know this will break Linux compatibilty, but anyone is welcome to take what I have and make it into a Linux driver. The reason for this immediate change is that I have some software in Windows that I'm using to edit patches. This allows me to selectively view the effect of certain MT-32 parameters. I figure if I can pipe the same stuff through a driver, I can then compare the output from both the emulator and MT-32 more directly. From here, I should be able to move rather quickly in getting more accurate emulation.
Yay.