Wonder if this thread from the other day will be useful:

SC-55 mkII "No Instrument" Message With canyon.mid and passport.mid

someperson42 wrote on 2021-05-26, 20:36:

I thought GS was a superset of GM, but given the way my SC-55ST is behaving, that doesn't seem to be the case? When I first power it on, or when I send a GS reset to the device, things like passport.mid are missing parts (most noticeable about 24 seconds in): https://1drv.ms/u/s!AlgWZf6mlC62zDcFvc3WNvmX-TWs?e=QyS6pD

If I send a GM reset to the device, then passport.mid plays as it should. Is this normal behavior? If so, can someone explain what's going on here?

Yeah, the thread referenced by Pierre is about the same problem. It seems everyone listens to canyon.mid and passport.mid nowadays 😀
To answer your question: Your version of passport .mid sends Bank Select + Program Change messages that are not valid on your SC-55. Invalid bank+program combinations result in silence on Roland gear (except the very first SC-55 that supported capital tone fallback). When your SC-55 receives a GM reset it disables the receiving of Bank Select messages fully so it always uses the default 128 GM compatible programs (capital tones) thus no invalid combinations are possible anymore.
GS reset enables the receiving of bank select messages in order GS specific variation banks to be available. But this has the risk of receiving invalid bank+program combinations sent by different Midi files.

Falcosoft wrote on 2021-05-26, 22:11:

Yeah, the thread referenced by Pierre is about the same problem. It seems everyone listens to canyon.mid and passport.mid nowadays 😀
To answer your question: Your version of passport .mid sends Bank Select + Program Change messages that are not valid on your SC-55. Invalid bank+program combinations result in silence on Roland gear (except the very first SC-55 that supported capital tone fallback). When your SC-55 receives a GM reset it disables the receiving of Bank Select messages fully so it always uses the default 128 GM compatible programs (capital tones) thus no invalid combinations are possible anymore.
GS reset enables the receiving of bank select messages in order GS specific variation banks to be available. But this has the risk of receiving invalid bank+program combinations sent by different Midi files.

Well, it seemed like a convenient way to test that my gear was working 😀

Anyway, that makes a lot of sense. Are there a lot of MIDI tracks that have this problem? I'm also curious if there are fixed versions of these tracks available somewhere.

someperson42 wrote on 2021-05-27, 19:51:

Well, it seemed like a convenient way to test that my gear was working 😀

Anyway, that makes a lot of sense. Are there a lot of MIDI tracks that have this problem? I'm also curious if there are fixed versions of these tracks available somewhere.

1. With your SC-55 you can meet such problematic Midi files since all the GS Midis that were written specifically for later SC devices (SC-88, SC-88Pro, SC-8820 etc.) can contain Bank+Program combinations that do not exist on SC-55. Also there are Midi files in the wild that are simply badly programmed. Even games can have such problems.
XG Midi files usually do not cause such problems since according to XG best practices before the XG reset a GM reset also should be placed.
Look for 'capital tone fallback' on Vogons for more info.

2. The versions of canyon.mid/passport.mid that can be found on Win3.1 do not have this problem (but have the drum on channel 16 problem instead. Read the thread referenced by Pierre). Earlier I posted these versions in my FSMP topic (MSMIDS.zip):
Re: Falcosoft Soundfont Midi Player + Munt VSTi + BassMidi VSTi

Falcosoft wrote on 2021-05-27, 20:31:

1. With your SC-55 you can meet such problematic Midi files since all the GS Midis that were written specifically for later SC d […]
Show full quote

1. With your SC-55 you can meet such problematic Midi files since all the GS Midis that were written specifically for later SC devices (SC-88, SC-88Pro, SC-8820 etc.) can contain Bank+Program combinations that do not exist on SC-55. Also there are Midi files in the wild that are simply badly programmed. Even games can have such problems.
XG Midi files usually do not cause such problems since according to XG best practices before the XG reset a GM reset also should be placed.
Look for 'capital tone fallback' on Vogons for more info.

2. The versions of canyon.mid/passport.mid that can be found on Win3.1 do not have this problem (but have the drum on channel 16 problem instead. Read the thread referenced by Pierre). Earlier I posted these versions in my FSMP topic (MSMIDS.zip):
Re: Falcosoft Soundfont Midi Player + Munt VSTi + BassMidi VSTi

I see. Sounds like there's no perfect solution here (other than replace my SC-55 with an earlier variant), which is a shame. I wonder why the "capital tone fallback" feature was removed.