PCEngineFans.com - The PC Engine and TurboGrafx-16 Community Forum
NEC PC-Engine/SuperGrafx => PC Engine/SuperGrafx Discussion => Topic started by: sunteam_paul on February 23, 2011, 06:40:32 AM
-
I thought this would be an interesting test to do - comparing hardware sound and emulation (in this case Magic Engine, which seems to get much criticism in this department).
I picked 3 completely random Hueys from my drawer that had easy to access music and did some recordings.
LISTEN HERE (http://soundcloud.com/peelfruit/pc-engine-hardwave-vs/s-LVpxl)
Please note that there will be obvious differences in bass and treble due to the shoddy way I very quickly did the recordings, so take that into account!
-
It's kind of hard to compare if some of the recordings are "shoddy". But so far I like the second example of each segment because it sounds brighter and less muddy.
-
Of the 3 primary emulators used for gaming right now (Magic Engine, Ootake, and Mednafen/PCEjin (Mednafen w/ GUI and full re-recording support), Magic Engine kinda has the worst sound emulation. It gets sounds wrong and has poor volume balance between PSG, ADPCM, and Redbook audio.
I would like to hear a good comparison between these three emulators and original hardware, though. That would be pretty rockin'.
-
The brighter ones is Magic Engine and the softer one is the real system (non 6280A). I can hear the tell tale DAC pops in the real version ( and I could see it when I zoomed in at the start of some parts). SGX and Core Grafx 1 have the better sound chip (IIRC the PAL only TurboGrafx 'no 16' has the better sound chip too, but it's 50hz so that sucks) since they lessen those pops to almost nill.
Why are your real system recordings filtered like that? Mine don't sound like that. If anything, they let more noise through (too much high frequencies to be honest).
-
Interesting, I never knew the SGX actually had a different sound chip than the others. I've always said my SuperGrafx has better bass than my Duo, but I thought perhaps it had something to do with IFU-30 amplifying the audio out or something.
Is there any possibility the "other" chip in the SGX/CG1 has better bass frequencies than the chip the Duos use?
-
poor volume balance between PSG, ADPCM, and Redbook audio.
You can manually adjust these can't you? I'm pretty sure I have.
Why are your real system recordings filtered like that? Mine don't sound like that. If anything, they let more noise through (too much high frequencies to be honest).
It's just recorded straight from the TV headphone socket. If I do another one I'll alter the TV settings to make it clearer.
-
Interesting, I never knew the SGX actually had a different sound chip than the others. I've always said my SuperGrafx has better bass than my Duo, but I thought perhaps it had something to do with IFU-30 amplifying the audio out or something.
Is there any possibility the "other" chip in the SGX/CG1 has better bass frequencies than the chip the Duos use?
Nah, the SGX/CG1 '6280A' chip just fixes the pops in the DAC for large volume changes (or turning on/off per audio channel). The audio itself coming out of the chip is pure digital looking beauty (perfectly square edged high frequency rise/fall steps on the scope). Then it goes into an amp and some other analog circuitry that filters here and there. I think my SGX has a little better bass in chip audio than my Duo, but neither compare to the unfiltered bass of mednafen emulation. I'd like to mod my SGX someday to those bass frequencies back. TmEE is the only person I know that completely redid the audio lines, amp, etc to a PCE/TG16 system. I'd be interested in seeing what kind of bass frequencies range he got out of it (without making a bass boosting amp).
I do know that the SGX audio output is inverted. I guess the Core Grafx is as well and that's why Ootake has inverted output for its emulation ('cause the original emulation author Ki based his RE'ing on the CG1). It would technically make some chip audio/effect sound a little different when played against CD audio or ADPCM simultaneously. But I seriously doubt you could even hear the difference. It'd have to be something like a phase differencing effect of a fluke to notice something like that.
sunteam_paul: Some games/tracks are more forgiving than others for the chip audio emulation. Magic Engine is still probably better than the winamp HES plugins. They tend get the noise channel (and levels) wrong.
-
I can hear the tell tale DAC pops in the real version
Yeah, that's the same thing I noticed. The real hardware also just sounds more crowded in general, which is typical of any low-end hardware mixing/muxing multiple sounds/streams.
The most objective way I can think of to record this would to record the line level output from the PCE (skip the TV entirely) into a PC for recording, and to do the same with the EMU. If the EMU is going straight to disc, that's really unfair. The real hardware has already been though a mile of crap (DAC, line level amp, cable) by the time you recorded it (two miles, if it went through your TV). The EMU should be recorded from the analog out of the PC making the sounds to another PC for recording (or some other method that would allow for an analog capture).
-
I'm going to try something like that now and see how it goes.
-
Pumping my old PC throught the TV and recording didn't work as planned so on to plan B:
Here's another collection of random tuneage. (http://soundcloud.com/peelfruit/pc-engine-audio-comparison/s-5BKoq)
I've tried my best to match the levels between the recordings, so they should be fairly comparable, although in my mind it's not a test of which is clearer or has more bass (as fiddling with the audio settings on a TV or amp can radically change the sound anyway) but how accurate the actual sounds are.
While it will never be 100% perfect, this test surprised me really. I was expecting a lot more of a difference and I could safely say if you played an emulated track to me out of the blue I wouldn't be able to tell if it was from real hardware or not. I certainly don't think ME deserve the bad rep it has. No doubts there are some tracks from other games that show more variation, but everything I tried so far is pretty damn good, if not excellent.
-
When I listen to PCE music from ME or Winamp, it always sounds unusual in various ways. They both seem to do a poor job with samples and system generated percussion types of sounds. Quite often, the overall tracks or particular channels/instruments sound very different and it feels broken. It might not sound like there is anything wrong to someone who isn't familiar with the music on real hardware. But it really stands out to me with soundtracks/games I listen to all the time.
-
I can't hear many differences in the second recording, though it probably doesn't help that my computer speakers are utter shite.
-
In an old thread, Kaminari posted a short clip comparing PSG in one of the Tengai Makyo(?) games, IIRC.
The differences between ME and real hardware were strong.
This leads to the obvious question: what is the best (or least horrible) way to listen to .hes rips?
-
I'd like to mod my SGX someday to those bass frequencies back. TmEE is the only person I know that completely redid the audio lines, amp, etc to a PCE/TG16 system. I'd be interested in seeing what kind of bass frequencies range he got out of it (without making a bass boosting amp).
I run the audio out through a Boss EQ pedal sometimes. You can do some fun things with it.
China warrior sounds great with the mids and highs scooped way down and the bass kicked up.
... and then you can do some more fun things with other pedals. Bonk sounds pretty cool with some thick chorus and delay added in.
Its a fun way to experiment with the sound of the console.
As far as emulators go, Ootake and Mednafen sounded closer to real hardware than Magic Engine when Insanity was being tested. Magic Engine was very brittle.
-
Here's one just for Arkhan: Insanity (http://soundcloud.com/peelfruit/pc-engine-music-comparison/s-650Kq)
I think this is the most obviously different of all the tracks I've tried so far and although subtle, can clearly be heard when compared side by side.
-
For added lols, do a recording of the chiptunes using MAGIC SYSTEM.
That stuff is a mess
-
sunteam_paul: Some games/tracks are more forgiving than others for the chip audio emulation. Magic Engine is still probably better than the winamp HES plugins. They tend get the noise channel (and levels) wrong.
There's a modified Nezplug+ floating around that claims to rectify some of the noise channel problems.
-
poor volume balance between PSG, ADPCM, and Redbook audio.
You can manually adjust these can't you? I'm pretty sure I have.
You can adjust these, but I prefer to load a game and go. No other emulator I've used has this problem. I can't imagine why the default volume settings would be so out of alignment.
-
poor volume balance between PSG, ADPCM, and Redbook audio.
You can manually adjust these can't you? I'm pretty sure I have.
You can adjust these, but I prefer to load a game and go. No other emulator I've used has this problem. I can't imagine why the default volume settings would be so out of alignment.
That's hardly a major issue. Do it once and it's done. Otherwise that would be like complaining your TV isn't very good because you don't want to adjust the brightness and contrast from the factory defaults.
-
I THINK I remember having had slightly different results with different games, too, though maybe the poor sound emulation just was throwing me off.
-
Paul, if there any games you think are good examples, Ootake can record WAV output so I should be able to capture what its sound emulation sounds like if you have something you want me to sample for you.
-
Ootake does nice PSG emulation!
Sounds pretty close to the real hardware, based off Insanity tests.
-
Here is another test
THE KUNG FU (new upload) (http://soundcloud.com/peelfruit/the-kung-fu-all/s-eIz9p)
In this test are the three primary PC Engine emulators (Magic Engine, Mednafen and Ootake) plus real hardware but NOT IN THAT ORDER!
I'm not telling you which is which so a gold star to anyone who can work it out.
-
Oops, looks like we can't find that page!
-
Oops, looks like we can't find that page!
Oopsie. Fixed!
-
Oops, looks like we can't find that page!
Oopsie. Fixed!
I'm on the page, but can't play it (iPhone & laptop).
-
Oops, looks like we can't find that page!
Oopsie. Fixed!
I'm on the page, but can't play it (iPhone & laptop).
Try again. I just did it with no problems (laptop running Linux, not logged in). Occasionally I find I have to refresh once and play again.
-
Here is another test
The Kung Fu (http://soundcloud.com/peelfruit/pc-engine-the-kung-fu/s-QuZj0)
In this test are the three primary PC Engine emulators (Magic Engine, Mednafen and Ootake) plus real hardware but NOT IN THAT ORDER!
I'm not telling you which is which so a gold star to anyone who can work it out.
How about at least posting what version # of the emulators you used, also what real system you used as well. I'm pretty sure I know what the real system is. One of the recordings sound very wrong in the opening part. Not sure what EMU it is, though.
-
Just an FYI, but Mednafen 0.9.16 has improved accuracy(in the digital portion, not the analog filter portion) PSG emulation compared to 0.8.x and earlier, and a setting "pce.psgrevision" to control which PSG revision to emulate("huc6280", "huc6280a"(default), "enhanced" -- read the documentation for more details). Many games are less clicky with HuC6280A, but there's one game(can't remember the name off hand) that sounds fine with HuC6280 but becomes a clicky mess with HuC6280A.
-
How about at least posting what version # of the emulators you used, also what real system you used as well. I'm pretty sure I know what the real system is. One of the recordings sound very wrong in the opening part. Not sure what EMU it is, though.
That opening part that was way off made me literally LOL. I had to rewind and listen again so I could keep laughing at it.
It definitely isn't Ootake, or the Magic Engine I have on this machine.... and my Mednafen doesn't do it either.
HMMMM....thats odd.
Paul, explain!
-
Well, most people complain about HES emulation over-emphasizing the noise channel (thus the percussion), so I'm tempted to choose the odd sounding one as the actual hardware, but that's too easy, so I plugged my Core into the TV and broke out The Kung Fu. Over the puny mono speakers of my TV the noise channel is actually pretty strong (almost dominant, just like the emulators). I also just tested the ROM in Ootake, on all three different modes of PSG filtering.
The 3rd recording is the odd one out and is definitely not original hardware or any of Ootake's PSG modes. I would guess Magic Engine because I think it has the least accurate sound emulation, but I don't know for sure.
The first, second, and fourth recordings all sound pretty similar, actually. It's difficult, at least from the recording, to make out any significant differences between them. I'm not comfortable guessing which is which. I would say that PCE PSG emulation is currently sufficiently accurate that you'd need good equipment and even better ears to really have reason to complain.
It also makes me wonder why anyone would bother, any longer, to pay money for Magic Engine, with such excellent free alternatives available. Yeah, the UI is nice, but it's also custom and doesn't make use of any of the common Windows widgets, which is rather pointless these days.
*edit*
Also, since I'm using a Core I, does that mean I have the "more awesomer" sound hardware that the SGFX supposedly uses?
-
How about at least posting what version # of the emulators you used, also what real system you used as well. I'm pretty sure I know what the real system is. One of the recordings sound very wrong in the opening part. Not sure what EMU it is, though.
I just checked and one of the emulators I used was a little out of date. For fairness, I have re-recorded it and uploaded it again. You may or may not notice any difference:
THE KUNG FU (again) (http://soundcloud.com/peelfruit/the-kung-fu-all/s-eIz9p)
I'll list what was where when people have had some more time to listen, discuss and guess what is what.
-
Either there's a lack of ability to really distinguish most of these, or there's a whole lot of apathy about this topic.
-
OK, here's the listing in order *drum roll*:
1. Mednafen
2. A real PC Engine Duo-RX
3. Ootake
4. Magic Engine
-
What settings did you use on Ootake when you recorded that? I ask because I checked with the latest version (just downloaded it) and it doesn't sound like your recording.
-
What settings did you use on Ootake when you recorded that? I ask because I checked with the latest version (just downloaded it) and it doesn't sound like your recording.
Defaults, also freshly downloaded.
The only thing I have done with the emulated sounds is to drop the treble to better match the output from the hardware.
-
I think that partially explains it. By default, Ootake does some weird "high quality PSG" stuff which reduces the treble a little. In fact, the mixing on Ootake overall is a little light on treble. If you further knocked off some treble that would definitely skew the results a little. I tried changing the PSG settings in Ootake but even on the lightest setting it's still a hair low in the treble department.
Interesting. I hadn't noticed this effect at all until you created this recording. It's also interesting, because I still find I like the way it sounds and find it to be very close to the sound of my PCE Core connected to my TV. I know the author of Ootake will take emails about his emulator. I'm tempted to send him a link to the audio file and see what he thinks.
-
What settings did you use on Ootake when you recorded that? I ask because I checked with the latest version (just downloaded it) and it doesn't sound like your recording.
Defaults, also freshly downloaded.
The only thing I have done with the emulated sounds is to drop the treble to better match the output from the hardware.
Doesn't that partially defeat the purpose of the comparison?
-
What settings did you use on Ootake when you recorded that? I ask because I checked with the latest version (just downloaded it) and it doesn't sound like your recording.
Defaults, also freshly downloaded.
The only thing I have done with the emulated sounds is to drop the treble to better match the output from the hardware.
Doesn't that partially defeat the purpose of the comparison?
How so? The real hardware was recorded via a TV which naturally produces a less clean sound than the emulators. Because I don't have a direct pipeline into my Duo's sound chip, adjusting the treble to match the conditions of the real hardware recording makes the comparison more valid than less.
-
I wouldn't say that. Just because you altered all the records, doesn't make them or it any more valid. The idea was fun, but I was under the impression you weren't messing with/altering anything.
-
I could just have easily turned up the treble on my TV when I recorded the real hardware and ended up with a different recording. I had to create a comparison where the bass and treble in all recordings were as equal as possible. This isn't what I would call messing around with it - if it was then I can make real hardware sound nothing like real hardware by fiddling with the equaliser on my amp. It's missing the point really.
-
Why did you have to record the Duo's audio via the TV? Can't you just use a simple adapter to run the audio line into your computer? Then you wouldn't have to do any adjustment of any kind.
-
Why did you have to record the Duo's audio via the TV? Can't you just use a simple adapter to run the audio line into your computer? Then you wouldn't have to do any adjustment of any kind.
I couldn't find the right leads. The emus would still have the advantage in clarity though, unless you ran them out the PC and back in via a cable...but I think that's going to far for what is essentially a simple test of the sound emulation itself, rather than the quality of the audio signal (which still wouldn't be an absolutely fair test). Besides, my PCE sounds like what comes out of my TV, so that's accurate to what I hear from real hardware when I'm playing.
-
OK, I finally listened to this, several times, and the differences between 1, 2 and 4 are rather subtle. #3 was the only obvious standout (weakest).
I actually am quite interested in this topic (sound emulation), so I would love it if you made more comparisons, Paul :)
Here's an an old thread about PCE vs. EMU sound (http://www.pcenginefx.com/forums/index.php?topic=1006.msg7751#msg7751), and I wish the sound files weren't broken links because the comparison was quite distinct.
Carry on...