The OPL3 doesn't play MIDI natively (it also doesn't do the digital playback that lets you play the .wav files, that's the Soundblster DAC). Normally an SB16 routes MIDI to the joystick port.

You can use software to play MIDI over OPL3 (badly), such as:
http://dosmid.sourceforge.net/

OPL doesn't support MIDI by default.. however there were drivers that allowed you to play MIDI through the OPL synth.

See here:
http://www.midiox.com/index.htm?http://www.mi … x.com/jsoft.htm

http://cd.textfiles.com/audio11000/MSDOS/SNDB … 13/MIDI-DRV.TXT

Also pretty sure that Windows 3.1 included an FM midi driver... or am I wrong?

cyclone3d wrote:

Also pretty sure that Windows 3.1 included an FM midi driver... or am I wrong?

I think you're right. I seem to recall that my Sound Galaxy card also included a Voyetra SAPI (or was it VAPI?) MIDI driver for Windows that sounded much better than the standard Windows 3.1 driver.

In Windows, you select "FM synthesis" for MIDI and Windows will do the rest. The media player can play MIDI files using the OPL3 then with no additional drivers or software required. That works since Win3.1 up to XP at least.

derSammler wrote:

In Windows, you select "FM synthesis" for MIDI and Windows will do the rest. The media player can play MIDI files using the OPL3 then with no additional drivers or software required. That works since Win3.1 up to XP at least.

What about in DOS ?

Intel486dx33 wrote:

[...]

What about in DOS ?

If you read my earlier post instead of ignoring it you might find something relevant...

The Creative "PLAY.EXE" utility can play .MID files in DOS IIRC, but it requires a full install of the SB16 drivers

IMO, the only reason to use an OPL to play MIDI is if you hate polyphony.

SirNickity wrote:

IMO, the only reason to use an OPL to play MIDI is if you hate polyphony.

That's complete nonsense. The OPL3 has 18 channels and unlike the 16 channels of a GM device, the OPL3 can play these at the same time - always! FM synthesis is not based on voices, and hence does not use multiple voices for instrument sounds. It can never run into a polyphony limit as GM devices do.

derSammler wrote:

That's complete nonsense. The OPL3 has 18 channels and unlike the 16 channels of a GM device, the OPL3 can play these at the same time - always!

Sorry to say, but your correction is even more nonsense. GM devices also can use all 16 channels simultaneously. Polyphony limit of early GM/GS devices (like SC-55) is independent from channel usage. Where do you get your information from?
https://en.wikipedia.org/wiki/General_MIDI

Criterion - Requirement Voices - Allow 24 voices to be available simultaneously for both melodic and percussive sounds (alterna […]
Show full quote

Criterion - Requirement
Voices - Allow 24 voices to be available simultaneously for both melodic and percussive sounds (alternatively, allow 16 melodic and 8 percussive voices). All voices respond to note velocity.
Channels - Support all 16 channels simultaneously, each assignable to different instruments. Channel 10 is reserved for percussion. Support polyphony (multiple simultaneous notes) on each channel.
Instruments - Support a minimum of 128 MIDI Program Numbers (conforming to the GM 1 Instrument Patch Map) and 47 percussion sounds (conforming to the GM 1 Percussion Key Map).

derSammler wrote:

FM synthesis is not based on voices, and hence does not use multiple voices for instrument sounds. It can never run into a polyphony limit as GM devices do.

Not exactly. OPL 'channels' actually can be considered as voices. Even one Piano instrument on one Midi channel can get polyphony limit when many notes are played, especially when sustain controller is also used. OPL3 can produce only 18 voices/notes simultaneously regardless of used Midi channels, and only if 2-op instruments are used. In case of 4-op instruments even fewer can be used. That's fewer than GM requires (24 voices).

This misconception has already been discussed here anyway:
Converting MIDI to OPL3/Adlib

MID-format player for Adlib device ?

jesolo wrote:

cyclone3d wrote:

Also pretty sure that Windows 3.1 included an FM midi driver... or am I wrong?

I think you're right. I seem to recall that my Sound Galaxy card also included a Voyetra SAPI (or was it VAPI?) MIDI driver for Windows that sounded much better than the standard Windows 3.1 driver.

Yes, but the built-in driver was meant for OPL2 only (though still works with OPL3, too).
There were a few other MIID->OPLx drivers around at the time, if memory serves..

Falcosoft wrote:

derSammler wrote:

That's complete nonsense. The OPL3 has 18 channels and unlike the 16 channels of a GM device, the OPL3 can play these at the same time - always!

Sorry to say, but your correction is even more nonsense. GM devices also can use all 16 channels simultaneously. Polyphony limit of early GM/GS devices (like SC-55) is independent from channel usage. Where do you get your information from?
https://en.wikipedia.org/wiki/General_MIDI

Actually, both of you might be right. There are GM devices out there that can't play two voices on separate channels exactly at once.

yawetaG wrote:

Falcosoft wrote:

derSammler wrote:

That's complete nonsense. The OPL3 has 18 channels and unlike the 16 channels of a GM device, the OPL3 can play these at the same time - always!

Sorry to say, but your correction is even more nonsense. GM devices also can use all 16 channels simultaneously. Polyphony limit of early GM/GS devices (like SC-55) is independent from channel usage. Where do you get your information from?
https://en.wikipedia.org/wiki/General_MIDI

Actually, both of you might be right. There are GM devices out there that can't play two voices on separate channels exactly at once.

According to GM specification such devices are not GM compatible devices. GM devices have to fulfill the above requirements. Using 16 channels simultaneously is not an optional feature of GM, it's mandatory.
So there are no GM devices out there that can't play two voices on separate channels since such devices are not GM devices by definition.

Falcosoft wrote:

According to GM specification such devices are not GM compatible devices. GM devices have to fulfill the above requirements. Usi […]
Show full quote

yawetaG wrote:

Falcosoft wrote:

Sorry to say, but your correction is even more nonsense. GM devices also can use all 16 channels simultaneously. Polyphony limit of early GM/GS devices (like SC-55) is independent from channel usage. Where do you get your information from?
https://en.wikipedia.org/wiki/General_MIDI

Actually, both of you might be right. There are GM devices out there that can't play two voices on separate channels exactly at once.

According to GM specification such devices are not GM compatible devices. GM devices have to fulfill the above requirements. Using 16 channels simultaneously is not an optional feature of GM, it's mandatory.
So there are no GM devices out there that can't play two voices on separate channels since such devices are not GM devices by definition.

GM1-rev4.pdf

Going by that document, quite a lot of "GM" devices are not actually GM because they differentiate from the specification in one way or another. The worst reasons to offend would be:
- Percussion can be played on another MIDI channel than channel 10 because drum sets can be assigned to different channels (includes many Roland devices).
- Not all Control Changes implemented (Casio, Kawai's simpler modules).
- No reaction to pitch bend for some instruments and often drum sets.
- Use of GM sound set on keyboards that lack MIDI in connector.
- Non-standard default volume setting: "64" instead of "100".
- "General MIDI" sound set that is numbered 1-128 and has the 16 instrument groups, except that the instruments provided don't follow the defined format.

Curiously, all of the MIDI modules that I own that have these quirks bear the "GM" logo and the required copyright notice in the manual, so it looks like the MMA did not look too hard at what modules could and could not do...

yawetaG wrote:

Going by that document, quite a lot of "GM" devices are not actually GM because they differentiate from the specification in one way or another. The worst reasons to offend would be:
- Percussion can be played on another MIDI channel than channel 10 because drum sets can be assigned to different channels (includes many Roland devices).

That's documented GS/XG extension. In GM mode (after a GM reset) all GM compatible Roland/Yamaha devices use channel 10 as Drum channel and all other channels as melodic by default.

- Not all Control Changes implemented (Casio, Kawai's simpler modules).

You could have been more specific about what controllers are not supported by these modules. Actually GM specification demands rather few controllers to be supported. Namely:

1 Modulation
7 Volume
10 Pan
11 Expression
64 Sustain
121 Reset All Controllers
123 All Notes Off

So e.g. even Reverb and Chorus support is not required by GM.

- No reaction to pitch bend for some instruments and often drum sets.

I do not see in the specification where required Drum/Percussion pitch bend support is mentioned.

But back to the point:
What MMA approved GM modules do you have/know that do not support simultaneous usage of all 16 channels?
Of course there can be such undocumented bugs and quirks in some modules. The question is if such rare quirks of some specific devices justify the generalized claim of derSammler that:
'The OPL3 has 18 channels and unlike the 16 channels of a GM device, the OPL3 can play these at the same time.'
I do not think so.

The SC-55, for example, was 24-voice polyphonic, and 16-part multitimbral. I'm sure there are plenty of sounds that use more than one voice per note, but compared to OPL, you're still going to have polyphony for days.

IMO, OPL2 sounds like garbage, and doesn't have enough complexity to simulate the (small G) general MIDI instrument set very well. OPL3 with a Super-SAPI driver is at least a little more interesting to listen to, but you can pretty much wipe out its polyphony with two chords on a piano track because of decay envelopes. Especially assuming the MIDI driver is going to reserve percussion channels (whether they're used or not in a given song.)

Even in OPL2 mode, canyon.mid is a glitchy mess. In OPL3 mode, just about every MIDI file I've ever played back has robbed notes like they were sitting unattended on the front porch in Detroit.