Early Sierra Games and the CM Series

[Great Hierophant]

The following early SCI-0 Sierra games exhibit odd sounds when playing with certain Roland MT-32 compatible modules:

King's Quest IV
Space Quest III
Leisure Suit Larry Goes Looking for Love in Several Wrong Places
Police Quest II
Hoyle's Classic Card Games: Volume I
Silpheed (not an SCI game, but from the same time period.)

Later SCI games sound fine I believe that the extra, often annoying and discordant sounds should only be present when the user is using a CM-64, CM-500 or a CM-32P (with a MT-32 or CM-32L.)  They will not be heard with a MT-32, MT-100, CM-32L or a LAPC-I.  The reason why they are heard is because the drivers for these early games must be inadvertently using parts 11-16.  Whateveris getting sent to part #2 is getting sent to part #11 and so on.  

If you recall an MT-32 compatible device only uses parts 2-9 for tones and part 10 for rhythym.  The CM-32P was designed to accompany the MT-32 or CM-32L and uses parts 11-16.  The CM-64 is a CM-32L + CM-32P in one module, nothing more or less.  The CM-500 is a CM-32L and uses the GS portion of the module to emulate the CM-32P parts in its mode B.  

How do I know this?  Well, if you send a sysex message to the CM-500 in mode B to turn parts 11-16 off, the odd, disharmonious sounds disappear, resulting in perfect MT-32 based playback.  Naturally, the same reasoning should apply to the CM-64, which I do not possess.  It makes no sense as applied to a CM-32L or LAPC-I, however, because they do not use the extra parts.  

The most important question is how widespread is the problem?  Considering that the price of an MT-32 or a CM-32L was high enough but the prices of the CM-32P and CM-64 approached ridiculous levels (for their PC purposes), it may be a bit more prevalent than one might think as companies never bothered to consider whether their code might not sound right on the "upgraded" or "enhanced" synthesis modules.

[Tom]

How do I know this? Well, if you send a sysex message to the CM-500 in mode B to turn parts 11-16 off, the odd, disharmonious sounds disappear, resulting in perfect MT-32 based playback. Naturally, the same reasoning should apply to the CM-64, which I do not possess. It makes no sense as applied to a CM-32L or LAPC-I, however, because they do not use the extra parts.

The most important question is how widespread is the problem? Considering that the price of an MT-32 or a CM-32L was high

The CM-500 defaults to playing both the L/A and GS synths, together.  You can either turn off the GS portion of the CM-500, or switch modes.  Roland often refers to these as 'Parts', but I would think the better nomenclature would be, "MIDI Channel"....because that's what they are -- MIDI channels.  The MT-32 only has 9, while almost every other MIDI device created since has 16.  

Even though Roland's newer synths state "32 channels", there are ONLY 16 MIDI channels...period.  By using multiple MIDI ports, you can multiply the number of available channels, in 16-channel increments across multiple ports.

This isn't really "a problem".  Disabling MIDI channels via SysEx has been part of MIDI since the beginning, so it's not unusual to do this.  In terms of Sierra MT-32 soundtracks, or any publisher's MT-32 soundtracks for that matter, they were primarily designed for the MT-32, and not for compatible devices.  A lot of publishers stated "Roland" support, in which they specifically mean, the MT-32.  I remember when I bought "LOOM" (LucasArts), and I had to send away for "Roland" drivers -- which were designed specifically for the MT-32.

I'm a firm believer in that --
If you want 100% MT-32 compatibility, buy a MT-32!  Not a CM-32L, or CM-500, or whatever.  (The LAPC is the only exception I'd make to this, as it trully is 100% compatible.)    All these other modules are great to have IN ADDITION to the MT-32, but I don't believe they can ever replace the real thing.

[Great Hierophant]

The CM-500 defaults to playing both the L/A and GS synths, together. You can either turn off the GS portion of the CM-500, or switch modes.  

That is Mode A, and it isn't very convenient to use for playing DOS games because you have to send a midi message to turn the GS "parts" off.  This wouldn't be too bad, but since MT-32 DOS games often customize their instruments you have to reset the device every time you play another game.  Otherwise, the game will likely not play properly.  While you could do this with sysex commands it's generally easier and quicker to push the power button.  Once you do that, you have to turn off the GS parts again.  

Mode B, which does not use the GS portion at all, is far more convenient if you are going to play alot of MT-32 files or games as I do.  But in order to play the Sierra games I mentioned correctly in Mode B, you have to turn off the PCM parts (11-16) which are emulated in Mode B.  Simply put, Mode B is more convenient to use for MT-32 files and games.  

If you want 100% MT-32 compatibility, buy a MT-32! Not a CM-32L, or CM-500, or whatever. (The LAPC is the only exception I'd make to this, as it trully is 100% compatible.) All these other modules are great to have IN ADDITION to the MT-32, but I don't believe they can ever replace the real thing.

Its a bit of overkill to connect an MT-32 and an CM-32L as the law of diminishing returns comes into its own in such a circumstance.  But the LAPC-I has the same enhancements over the MT-32 that the CM-32L has.  The LAPC-I and CM-32L both use the same PCM and Control ROMs.  There is no difference between the two when it comes to the LA synthesis.  Therefore, if the LAPC-I is 100% compatible with the MT-32, the CM-32L must also be 100% compatible with the MT-32 or neither are 100% compatible with the MT-32.  Essentially, for purposes of LA synthesis the CM-32L, the LAPC-I, the CM-64 and the CM-500 are the same module.

[Cloudschatze]

The phenomenon experienced with the CM-64 and CM-500 is due to the fact that the MT-32 and Adlib devices share the same soundtrack. Those "odd sounds" are the playback of MIDI data intended for an OPL synth.

MT-32 MIDI data is stored on channels 1 - 10
Adlib MIDI data is stored on channels 11 - 16

At the time of their release, a sixteen-channel, MT-32-compatible module did not even exist, thus, Sierra's original MT-32 driver was adequate. With the release of the CM-64 and CM-500 units, Sierra must have encountered the issue, and consequently updated their driver to disable/mute channels 11 - 16 upon startup.

If you were to edit the MT-32 driver for one of the games you've mentioned, you will notice that it states only MT-32 and LAPC-I near the top. You will need a driver which specifies MT-32, LAPC-I and CM-64, etc. Copy one of these newer MT32.DRV drivers into the game directory.

This effectively solves the problem, with Sierra games at least. I've not yet encountered this issue with any others.

[Great Hierophant]

Thanks for reminding me, I had heard that before but forgotten it.  That does make sense considering the Adlib's YM-3812 sound chip could be programmed for 9 channels or 6 channels and 3 percussion channels.  

If you were to edit the MT-32 driver for one of the games you've mentioned, you will notice that it states only MT-32 and LAPC-I near the top. You will need a driver which specifies MT-32, LAPC-I and CM-64, etc. Copy one of these newer MT32.DRV drivers into the game directory.

Wouldn't your solution will only work if you have a later version of the same game?  After all, if you copied an MT32.DRV, even with support for the CM-64 , from another Sierra game it would not sound correct because the .DRV file contains the custom instrument patches for the other game.

[Tom]

I didn't know that -- that the CM-32L is the same as the LAPC.  I don't ever remember seeing the CM-32L for quite some time after the LAPC was around, and I had only heard it compared to "the MT-32 on a card", and not the CM-32L on a card.

I've never been able to make my CM-500 behave exactly like the MT-32, either in Mode A or B.  There seems to be differences in overall balance with some songs, and with some of the timbres sounding a bit different.

"Overkill" is in the eye of the musician.  I never would consider using a real MT-32 as overkill, unless all I was interested in was using it to play games.  I probably look at this a bit differntly.  I would have never spent over $500 for the MT-32 back in 1988...just for playing games.  I buy musical devices primarily to make music with; using them for gaming is a wonderful perk!  By 1990, I had already accumulated a large library of MT-32 songs from many musicians all over the world.  I've found that many of these songs don't play back well on my CM-500.  This is another reason that for me, the 'real thing' is worth it.

[Great Hierophant]

From perusing the manuals, the most noticeable differences between the LA synthesis devices seems to be their default-power up values.  There are some minor differences in the charts of seemingly each module.  A pan setting here, a partial reserve there.  While these values may have drifted over time, I believe that the manuals were simply mis-typed and in need of errata sheets.  I doubt Roland would fiddle with the default settings when the module is almost infinitely customizable.  

Even if the settings are slightly different from module to module, those are just the start up values and as we all know the Roland MT-32 rewards attention and tweaking.  So, the question is what is so different about the MT-32 compared to its later descendants that affects the audio qualities of the synthesis?  

Using an MT-32 for gaming is not overkill, its very practical considering how difficult and expensive finding and buying other synthesizers can be and how hard it is to emulate Roland's LA synthesis with real accuracy.

[HondaSiR]

Using an MT-32 for gaming is not overkill, its very practical considering how difficult and expensive finding and buying other synthesizers can be and how hard it is to emulate Roland's LA synthesis with real accuracy.

What Tom probably meant was it was quite an overkill to buy an MT-32 just for gaming purposes back in the days when it was new since a module costs over $500.00 (1987-1990). Today if you buy an MT-32 in ebay, its not overkill since a unit costs just around $30.00 or so. Ironically, nowadays its the MT-32 compatibles that command astronomical prices like the LAPC-1 and CM-500.

[Cloudschatze]

Wouldn't your solution will only work if you have a later version of the same game?  After all, if you copied an MT32.DRV, even with support for the CM-64 , from another Sierra game it would not sound correct because the .DRV file contains the custom instrument patches for the other game.

The custom patches for the MT-32 are contained in a file called PATCH.001, which resides within the game's resource files. Substituting another MT32.DRV, even from another game, will not change this.

[Ari]

Didn't Sierra stop using channels 11-16 altogether with the newer games? They would need to have different soundtrack files for earlier GM games anyway, no? (before they started supporting the MT-32 just as an afterthought)

When I recorded the SQ1VGA and SQ4 MT-32 soundtracks, there wasn't any data on these channels anymore.

[Alistair]

Didn't Sierra stop using channels 11-16 altogether with the newer games?

No, but they stopped including them all in the one driver. The MT-32/GM drivers no longer had the Adlib content on them.

The Adlib drivers used 11-16, IIRC.

- Alistair

[Ari]

I meant use them in the same soundtrack files for the MT-32.
Or to makes things a bit clearer - use different soundtrack files for MT-32, GM and adlib, where the MT-32 and GM soundtrack files contain no data for channels 11-16.

[Zemus]

I've never been able to make my CM-500 behave exactly like the MT-32, either in Mode A or B. There seems to be differences in overall balance with some songs, and with some of the timbres sounding a bit different.

Tom, do you have any comparisons where you can hear the differences? I'd be interested to know how different my CM-500 sounds to the MT-32.

[Tom]

Yeah...several.  The first time I tried the CM-500, I played some KQ4 MIDI files and noticed the difference in sound, right away.  No patch bank being used, either.  I can't record anything right now; my CM-500 is presently sitting on top of my VCR, not hooked up to anything.  I plan on hooking it up to my music PC after I install that V-22 Interface Cloudschatze sent me.  When I have some spare time, I'll record a couple of examples.  (Didn't we do this before -- with LAPC vs. CM-500 playback?)

[Cloudschatze]

Yeah...several.  The first time I tried the CM-500, I played some KQ4 MIDI files and noticed the difference in sound, right away.  No patch bank being used, either.  I can't record anything right now; my CM-500 is presently sitting on top of my VCR, not hooked up to anything.  I plan on hooking it up to my music PC after I install that V-22 Interface Cloudschatze sent me.  When I have some spare time, I'll record a couple of examples.  (Didn't we do this before -- with LAPC vs. CM-500 playback?)

Okay, I'm intrigued. From my KQ4 experience, if you've set the CM-500 to Mode B, and have performed the MT32 driver replacement, the output should be nearly identical to that of the MT-32. The only discernable difference between the two that I've found is that the CM-500 produces a slightly "crisper" sound.

That, and Leisure Suit Larry 5's use of sound-effects not present in the MT-32.

[Great Hierophant]

Okay, I'm intrigued. From my KQ4 experience, if you've set the CM-500 to Mode B, and have performed the MT32 driver replacement, the output should be nearly identical to that of the MT-32. The only discernable difference between the two that I've found is that the CM-500 produces a slightly "crisper" sound.

That, and Leisure Suit Larry 5's use of sound-effects not present in the MT-32.

I have tried replacing the MT32.DRV of King's Quest IV from several later Sierra games that don't have the problem I discussed.  In every case, it seems that only either half the notes play at any given time (Sierra SCI-0 games without the problem), the music is awfully distorted (other Sierra games with the problem) or the game refuses to run (SCI-1 games).  Could you suggest an appropriate game's MT32.DRV to use?  

Leisure Suit Larry 5 obviously must be using the LAPC-I/CM-32L effects.

[Cloudschatze]

I have tried replacing the MT32.DRV of King's Quest IV from several later Sierra games that don't have the problem I discussed.  In every case, it seems that only either half the notes play at any given time, the music is awfully distorted or the game refuses to run.  Could you suggest an appropriate game's MT32.DRV to use?

It would have to be from another SCI0 game. Quest for Glory I, Codename: Iceman, King's Quest I:EGA, etc.  

Leisure Suit Larry 5 obviously must be using the LAPC-I/CM-32L effects.

Yep.  

[Great Hierophant]

The only applicable games would be as follows:

Leisure Suit Larry 3
Quest for Glory I/Hero's Quest
King's Quest I
The Colonel's Bequest
Codename: Iceman
Conquest of Camelot

Quest for Glory II will crash the games because it uses an SCI version inbetween 0 and 1.

[Alistair]

Or to makes things a bit clearer - use different soundtrack files for MT-32, GM and adlib, where the MT-32 and GM soundtrack files contain no data for channels 11-16.

That's what I said

- Alistair

[Tom]

MT32 driver replacement

? I didn't replace any drivers -- didn't know I had to back when I tried it.  

But anyway, I'm not talking about game musicm per se; I'm referring to MT-32 MIDI files I've accumulated over the years by other musicians.