The truth about waveblaster - the sequel

[WBman]

Impressive work, Salient, thank you very much! No matter if whoever prefers SC-55, WB or MU-2000, this is now pretty useful and nice comparison. I like many of the SC-55 versions too. Thanks.

[Salient]

Impressive work, Salient, thank you very much! No matter if whoever prefers SC-55, WB or MU-2000, this is now pretty useful and nice comparison. I like many of the SC-55 versions too. Thanks.

'
Thanks.
I find the MU2000 a bit more inconsistant than the SC55.

[Salient]

Recording request for Andrew also done.

Recorded on a real Wave Blaster 1, mounted on an external interface (home-made), no extra effects and device re-initialized before each new recording. Due to interference of my home-made interface, 50hz noise was removed using auto profiler in Audacity v1.3.

Here are the files: CLICK HERE

Note: SC-55 and MU2000 recordings will follow later, it's over half past 11pm and I really want to go to bed now

[Ari]

Of course I listened to both of them. I really didn't find much classic orchestral stuff there. And why to record songs like Orion's Belt Bar 1-3, I don't see much point. I tried quickly play back few tunes from the "complete game music" MIDI rip and it sounded quiet strange and buggy on WB. Then I tried to run the game on a DOS computer and voi-la, it sounded much nicer. Although still nothing great as these definitely are not much WB suitable soundtracks. I then tried to test other MIDI rips from here and started with Betrayal at Krondor - the game I know very well, as well as its music. And I was startled - it's completly wrong. For example instead of strings patch 48 playing in the game, it calls synth strings patch 50 + other instruments mistakes, there are also different volume balances. So I don't know but there's is something wrong with these rips. I'm not saying that with all the rips, but at least BAK rip sounds wrong and SQs also sound slightly better when I start the games itself under the DOS, at least the intro sequences.

Why should these themes sound any different under DOS and under Windows?
Could you perhaps somehow record the 2 with the WaveBlaster to show the differences?

[Salient]

To be honest, I find the recordings I made (under Windows) on the Wave Blaster 1 of the space quest 5 and 6 tracks quite alright actually (except maybe for the bar theme 1). If it is even better under DOS I would really like to hear the same tracks recorded from DOS.

[Salient]

I made some new comparison recordings.
The files are recorded on:
- Wave Blaster 1
- Yamaha MU2000
- Roland SC-55
- Roland SC-8820 native
- Roland SC-8820 in SC-88 map mode

The files are here.

[andrew603]

To be honest, I find the recordings I made (under Windows) on the Wave Blaster 1 of the space quest 5 and 6 tracks quite alright actually (except maybe for the bar theme 1). If it is even better under DOS I would really like to hear the same tracks recorded from DOS.

Who could have anticipated this?  Oh, I did.  Weeks ago.  

Yes, the SQ music is not bad on WB!  I still prefer SC-55/88 or Korg X5DR on these soundtracks, but for the most part the WB is quite acceptable here with only a few exceptions.  Even the ghastly electric guitars kinda sound OK as their own thing in "Roger's Place".

[WBman]

Why should these themes sound any different under DOS and under Windows?
Could you perhaps somehow record the 2 with the WaveBlaster to show the differences?

I now have connected some other gear for a while, so new recordings are unavailable at the moment, but I can record the DOS ver. of SQ within few days. But when I tested the SQ 5,6 and BOK MIDI rips recently, I recorded some BOK sketches showing differencies between real DOS performance and the QS MIDI rip of similar tune. In the principle it shows the problems and differencies of the SQ rips also and even of most MIDI rips in general.
This is the excerpt from original MIDI performance from DOS running Betrayal at Krondor:
http://sites.google.com/site/waveblastergm/Krondor-original.ogg

Look for the panning: strings-right, flute-left, horn-hard right, bass-almost center
And this is the QS MIDI rip "General MIDI" version:
http://sites.google.com/site/waveblastergm/Krondor-QS_GM_rip.ogg

At first listening it sounds alike but:
Strings are incorectly panned left, flute is left (as in the original) but also horn is incorrectly hard left. So much instruments layered to the left of course sound strange, especially the horn doesn't match. It can look like relatively minor errors, but in fact it worsens the overall balance and especially the stereo image. In other MIDI rips I was encountering similar problems. But it definitely isn't related to QS MIDI rips only, but almost all MIDI rips I have checked have more or less errors, sometimes rather imperceptible, sometimes very noticeable. That's why it's really better to record audio from the running game itself, preferably real DOS, or from unaltered ripped music file (typically XMI if possible) played back by a basic DOS player using the game's drivers. MIDI rips are almost always imperfect, as you can hear from the BOK example.

Yes, the SQ music is not bad on WB!

I don't like the WB version, it sounds not right to me. In some parts, it is even out of tune.

[Marten]

WBman,

As far as I am aware, all of Quest Studios' MIDIs are created in one of two ways, and neither of those ways is to "rip" the MIDI in the way you are suggesting has been done.  Tom, Ari, please correct me if I'm incorrect on this... but I distinctly remember Tom mentioning in the past that he does not rip the files.

The first method by which Quest Studios' MIDIs are created is to record the MIDIs directly from the game while the game is being played.  In this case, the digital data that is sent to the MIDI port is captured directly to a file, and by replaying this file at a later time and sending it to the appropriate device, the result will be identical to what you hear playing the game with the appropriate device originally.

The second method by which some of Quest Studios' MIDIs are created is to modify files created the first way, to convert them for General MIDI.  Tom and others (such as Ari) have done this manual conversion for a number of soundtracks that were originally intended only to be played on an MT-32.  For these soundtracks, there is no official General MIDI version, and no General MIDI device - not even the Sound Canvas that some people here laud with such high praise - will play a pure MT-32 soundtrack correctly.  The instrument mapping is entirely different.

Now... Krondor is a curious example that you've chosen.  As far as I've been able to tell by researching it, the game was produced in 1993 and should have supported both MT-32 and General MIDI.  But, some of the files on the Betrayal at Krondor Sound Canvas GM page, particularly contributions from Ari, say they were "converted for General MIDI".  I'm not entirely sure why Ari would have converted the files, this if the game already supports General MIDI.  Maybe he can explain...

By the way, could you please indicate which of the QS MIDI files you used for your Betrayal at Krondor sample?  I don't want to listen to every single one of the files on the site, plus your sample, to figure out which one you used.  Knowing which file you used is important, because then we can look ask Tom or Ari how that particular file was created... if it was an MT-32 to General MIDI conversion, there's certainly no guarantee of it sounding "correct" because the converter's bias on what he thinks sounds good, or what he thinks sounds closest to the original MT-32 recording, will influence the results.

[Ari]

Marten, you are essentially right. 99% (I daresay 100%) of the MIDIs on this site were made using one of the techniques you described. There are other methods for recording without ripping - such as using DOSbox's internal MIDI recording that simply logs all the MIDI commands that the game sends out to the sound device, however I am unaware of any recordings that were done this way and uploaded to the website.

Regarding 'Betrayal at Krondor', I'm not sure, but there appears to be some difference in the actual MIDI data between the MT-32 version and the GM version when choosing either one in the sound setup program for the game. It would seem that  the MT-32 version doesn't simply use sysex to remap the MT-32 instruments with the GM data to make the soundtrack MT-32 compatible, like in most Sierra soundtracks from later games (1992 and newer). Some of the older GM BAK recordings were actually recorded in MT-32 mode (one of those seems to be 'Krondor'). I will try to verify that.

I don't quite remember why we wrote "converted to GM" in some of my recordings. It might just be a misprint. I did do some remixing with some of the themes - mainly replacing the 'Standard' drum kit with the 'Orchestra' drum kit, but other than that, there was no "conversion" to GM.

SQ5 and SQ6 are a whole other ball game. These were recorded, AFAIK, entirely using the GM setup. This needs to be verified with Tom.

[Salient]

@WBMan:

Do you have this article and if so , can you scan it for me please?

Weksler, Mike & McGee, Joe. "CGW Sound Card Survey." Computer Gaming World Oct. 1993: 76-84

I am kind of a collector of all sorts of soundcard tests and what I have so far I've put here: Click to see

[WBman]

OK, the MIDI file of the sample above was Bak2gm.mid, and the nfo really says "Converted from the Roland MT-32 version", so you were right about this one. However I encountered problems and differences even with files that say "recorded directly using Sierra's GM sound driver". Here are the samples of such files:

File "Bakgm11a.mid" - Fight to the Death - Part One
Recording from DOS running game:
http://sites.google.com/site/waveblastergm/FighttotheDeath1-originalDOS.ogg
Recording from QS GM MIDI rip:
http://sites.google.com/site/waveblastergm/FighttotheDeath1-QSMIDIrip.ogg

The differences are obvious. In the MIDI rip - strange and quit percussion, strings somehow muddier, something wrong with french horn, like if it was muddied by doubling at similar channel or someting. The original DOS - more balanced, sharper percussion, strings and brass. Of course I always reset the device, so it can't originate randomly by an error.

File "Bakgm10.mid" - Riddles and Locks
DOS running game:
http://sites.google.com/site/waveblastergm/RiddlesandLocks-originalDOS.ogg
QS GM MIDI rip:
http://sites.google.com/site/waveblastergm/RiddlesandLocks-QSMIDIrip.ogg

Here, in the MIDI rip there's even some electric piano that shouldn't be there at all.

I checked all the possible settings that eventually could alter sound somehow when playing the MIDI rips, I repeatedly reseted the WB, tried other MIDI files for sure, and my result is that there are differences between the running game music and the MIDI rips in some BAK songs. You can hear it in these examples.

[Cloudschatze]

Weksler, Mike & McGee, Joe. "CGW Sound Card Survey." Computer Gaming World Oct. 1993: 76-84

http://vogons.zetafleet.com/viewtopic.php?t=21243


You may want to take Mike Weksler off of your list of "experts," WBman. According to the 1995 article also referenced in the link above, he states that the Wave Blaster II is "better than the original WAVE BLASTER." 

[WBman]

You may want to take Mike Weksler off of your list of "experts"

No reason at all. Everybody has the privillege to be mistaken sometimes, especially when reviewing bunch of new products (and the older ones only remembering from 2-3 years ago).
If you review a product, and after two years you get the product's sequel claiming to be an improvement, you tend to judge it as better at first, particularly if you don't have the original product for direct comparison. I have no problem with this. It's still a normal review, far above one unnamed "review" where the reviewer feels to be "ripped-off"

[Salient]

But what about the opinion of Bobby Prince?

I think his opinion does have some value at least.

About Bobby (Robert) Prince

[Cloudschatze]

You may want to take Mike Weksler off of your list of "experts"

No reason at all. Everybody has the privillege to be mistaken sometimes...

My, my... Isn't that presumptuous? 

But what about the opinion of Bobby Prince?

I think his opinion does have some value at least.

I'm at odds with him on the Ensoniq Soundscape DB rating. I haven't heard the ALR and Aztech boards (yet...), but I've done listening tests with everything else on his list, and the (1MB) Ensoniq is, in my opinion, the worst sounding DB of the lot.

[Marten]

Everyone is entitled to his or her own opinion, and there is no right or wrong in matters of personal taste.

I'm curious about the reason behind the stated differences between Bakgm10.mid and Bakgm11a.mid using the MIDI recording with the MIDI audio device, versus the digital audio recording direct from the game.  The only thing I can think to ask is whether the the tool being used to playback the MIDI is somehow causing the problem, like causing a MIDI buffer overrun.  But that would be a tool problem, not a MIDI file problem.

And, I'm sorry but I must emphasize again that Tom's files are not "MIDI rips", but "MIDI recordings".  This may seem to be a technicality to some people, but "MIDI rip" means that the MIDI data was extracted from the game data files, whereas a MIDI recording is a capturing of the MIDI data as the game sends it to the intended device, and saving a copy to disk.  The problem with ripping MIDI (versus recording MIDI) is that the game may apply changes to the MIDI data that is not present in the raw data - just as an example, the game could add additional reverb or shift the left/right balance as part of a game mechanic, while the original MIDI data might not contain those instructions.  So, if you playback a MIDI rip, you might miss special effects that would be present in the game.  Recording, as Tom does, preserves all of the MIDI instructions in a compact file that (with a correctly working playback tool) can be transmitted to your MIDI device to reproduce an authentic experience.

[WBman]

But what about the opinion of Bobby Prince? I think his opinion does have some value at least.

It's relative. For someone, for example, would be much more relevant to hear opinions of some more orchestral skilled guys like Eric Heberling, Dave Govett or Clint Bajakian. There are plenty of other factors too - except the basic subjectivity also how deep and long the cards were tested, if there were 5+ games worthfuly tested, or only few demo MIDI files etc. However I agree with Cloudschatze about odd Ensoniq 1MB rating. I have some other journal's sound card reviews (but unfortunately not in the English language) claiming some yet another different opinions.

@Marten: I'm too curious why they sound different. I understand that they are MIDI recordings and should sound fine. I think the MIDI player shouldn't be a problem, I tried the basic Win 98 WM player, DOS MIDI player, as well as XP WMP, and tested it with other MIDIs from various games which sounded fine. But if you open the bakgm11a.mid file in a sequencer, you can see the percussion is based on the note D1 (snare drum) and has very low velocity settings, but in the game itself it's clearly the note E1 and higher values. So I see the differencies in the file.

Btw. doesn't anybody know anything closer about this card mentioned in one review:
ALR Wave Express Module - it should have 6MB Kurzweil soundset - this is very promising as Kurzweil sounds are generally higly acclaimed. But I wasn't able to find any info at all, almost as if the card wouldn't exist. Or couldn't it be renamed?

[Ari]

As I tried to explain before, these differences could be the result of the fact that some of these files were recorded using the MT-32 driver. I will attempt to verify that.

[Cloudschatze]

Btw. doesn't anybody know anything closer about this card mentioned in one review:
ALR Wave Express Module - it should have 6MB Kurzweil soundset - this is very promising as Kurzweil sounds are generally higly acclaimed. But I wasn't able to find any info at all, almost as if the card wouldn't exist. Or couldn't it be renamed?

ALR Wave Express = Wearnes EFX-2000 = AVM Summit (DB in external enclosure)

Here is a photograph of one: http://wavetable.web.fc2.com/avm.htm