Befure testing on real hardware I just tested this on PC, but this is not a precise conversion I think... I think volume has trouble and is unbalanced. Inside the same player if I set s98 and VGM I've different result 🙁 It seems that YM2203 or YM2608 recognition is automated. Seems like that the generated s98 file is YM2608 compliant even if VGM is YM2203. Btw, this conversion works on real hardware... I've to record and compare vgm/emulator, recorded version to understand if the results is what I'm looking for. Thank you for the file.

bakemono wrote on 2022-12-17, 13:34:

You can try this program to convert from VGM to S98.

BTW, has anyone found a VGM rip from PC-88VA 神羅万象 (Shinra Banshou) ? It was in a video clip on nicovideo and sounded pretty good for a game from 1987 (?)

edit: that one had problems. Try this instead.

Now that I've tested a lot, what I've to say is: Thank you, Thank you, Thank you.

I've used this until now for silpheed soundtrack. There's a difference from joshw.info archive and vgmrips. and joshw.info archive in case of silpheed had one track missing.
So I've used your exe file to convert the missing soundtrack. VOlume is higher than the other s98 file but results was imho great.

This is the track, vgm to s98 converter:
https://youtu.be/LPueMEJO7aA

What i would really like to ask, if i can is 2 features: the first one: there's no loop point in s98 generated (while loop are inside VGM). In this case, i looped manually the track working on samples, nut Eve burst error are more than 40 tracks and so on... do you think it could be possible to create the s98 file with loop information? This would be relaly great.
Other request concern about multple YM2203 track (if there's 2 YM2203 inside the VGM it would be possible to create to file from 1 vgm (chip one and chip two))??

Really thank you for your help.

Aldo

Loop point should be doable. I'm not sure about doing multiple sound chips with version 1 format. Do you have an example .S98 file that uses two chips so I can see what the headers look like?

bakemono wrote on 2022-12-24, 10:48:

Loop point should be doable. I'm not sure about doing multiple sound chips with version 1 format. Do you have an example .S98 file that uses two chips so I can see what the headers look like?

Thank you, implementing loop point would be great (and set number of loop with something like -2 -3 absolutely fantastic).

What I was meaning concerning multiple chip, was not the s98 files, but the VGM:
https://vgmrips.net/packs/pack/legendary-wings-arcade
There are really few soundtracks, btw, in this vgm for example i can see 2xYM2203, what i intended was to create 2 -s98 files from a single VGM (one for each YM2203 chipset)
Btw there are really few soundtrack, so maybe this is unnecessary if it's an hard work.

Aldo

I see. Well the thing is that S98 version 2 is supposed to have its own support for 2 chips. I don't know about version 1 because I didn't a find a specification for it. But taking a VGM with 2 chips and splitting it into two S98 files with one chip each is also not too hard. Run the program twice, processing one chip at a time with switches (-one and -two)

Since I found a Shinra Banshou dump that I was looking for, as S98, I made an S98 to VGM converter as well...

Thank you.
Just downloaded. Now I'm working on some MT-32 OST, btw i'll try this isoon... and I'm going to upload Shinra Banshou OST on my channel asap (I need to show gratitude).
Thank you again.
Aldo
P.S.: loop?

Loop points should be working

bakemono wrote on 2022-12-27, 15:20:

Since I found a Shinra Banshou dump that I was looking for, as S98, I made an S98 to VGM converter as well...

s98 files are fot 2608 or 2203? Do 2203 version exists?
thank you.

RetroGC wrote on 2022-12-31, 21:01:

s98 files are fot 2608 or 2203? Do 2203 version exists?
thank you.

I found a PC-88VA version and another version but they are both YM2608.

bakemono wrote on 2022-12-27, 15:20:

I see. Well the thing is that S98 version 2 is supposed to have its own support for 2 chips. I don't know about version 1 because I didn't a find a specification for it. But taking a VGM with 2 chips and splitting it into two S98 files with one chip each is also not too hard. Run the program twice, processing one chip at a time with switches (-one and -two)

Since I found a Shinra Banshou dump that I was looking for, as S98, I made an S98 to VGM converter as well...

Thank you. They worked great. I finally started to upload Eve Burst Error, with real hardware recording.

https://youtube.com/playlist?list=PL_BhbJAAue … a5n6v6wWUW8dTKE

Hi. One of my Japanese friends told me that ssg output is too loud compared to a real pc-98 because usually ssg output of pc-98 was keeped lower. Maybe you could add a step-down ssg in your converter (maybe I could tipe ssg-2 ssg-5 etc) to arrive to determine an exact balance in output? First of all sorry for bothering. Than, thank you for your patience.

bakemono wrote on 2022-12-27, 15:20:

I see. Well the thing is that S98 version 2 is supposed to have its own support for 2 chips. I don't know about version 1 because I didn't a find a specification for it. But taking a VGM with 2 chips and splitting it into two S98 files with one chip each is also not too hard. Run the program twice, processing one chip at a time with switches (-one and -two)

Since I found a Shinra Banshou dump that I was looking for, as S98, I made an S98 to VGM converter as well...

Hi,
First fo all, sorry for bothering.
Can you make this software have exactly the same output f the VGM?
I'm doing some try with touhou 2, but about at 1 minut 24 seconds... i think it output a bad sounding output.
Here is the VGM I'm working on, it's a little bit different from the original, because thanks to an user from VGMRips, here it's mixed with reduced volume the SSG part.
Now I'm testing those files with kumatan MDPlayer software.
Thank you.
Aldo

RetroGC wrote on 2023-05-01, 17:58:

I'm doing some try with touhou 2, but about at 1 minut 24 seconds... i think it output a bad sounding output. Here is the VGM I' […]
Show full quote

I'm doing some try with touhou 2, but about at 1 minut 24 seconds... i think it output a bad sounding output.
Here is the VGM I'm working on, it's a little bit different from the original, because thanks to an user from VGMRips, here it's mixed with reduced volume the SSG part.
Now I'm testing those files with kumatan MDPlayer software.
Thank you.
Aldo

This VGM seems like a bad dump, as it is marked YM2203 in the header but the file contains YM2203 and YM2608 data mixed together. But I modified the program to ignore the extra data, so maybe it will be closer to what was intended.

What king of extra data contains? (music? Maybe this was intended for YM2203 and i don't know in what way become 2608?)

Thank you. I downloaded it from VGMrips from this link:
https://vgmrips.net/packs/pack/touhou-fuumaro … ec-pc-9801-opna
then I used a pyton script (vgm-edit.py) to reduce SSG volume, maybe that caused this kind of problems?
Then I needed to convert it again to s98 files to reproduce it in my hardware.

I've similar problems i think with PMD2S98: thanks to hoot archive, there's ton of .M files inside the Zip files. The zip file I'm attaching contains, according to Hoot Display an YM2203 OST.
I've converted some .M file (exactly the desire files for YM2203) to S98 (and then s.98 to VGM), but while inside Hoot, i can see the M file is intended for YM2203, after those conversion I've the S98 generated claiming it for 2608 (and in this way i can not reproduce it with hardware) and on MDPlayer they seem to be for 2608.

Thank you for your help.

Aldo

bakemono wrote on 2023-05-01, 21:20:

RetroGC wrote on 2023-05-01, 17:58:

I'm doing some try with touhou 2, but about at 1 minut 24 seconds... i think it output a bad sounding output. Here is the VGM I' […]
Show full quote

I'm doing some try with touhou 2, but about at 1 minut 24 seconds... i think it output a bad sounding output.
Here is the VGM I'm working on, it's a little bit different from the original, because thanks to an user from VGMRips, here it's mixed with reduced volume the SSG part.
Now I'm testing those files with kumatan MDPlayer software.
Thank you.
Aldo

This VGM seems like a bad dump, as it is marked YM2203 in the header but the file contains YM2203 and YM2608 data mixed together. But I modified the program to ignore the extra data, so maybe it will be closer to what was intended.

I'll attach here all the conversion. I'm always getting volume differences using the same player. All the s98 file I've attacched I've created with the new release of your converter.
Inside the Zip, the original VGM files from VGM rips, my edited files with mixed channel (thanks to vgm-edit.py).
Thank you for your help.
Aldo

RetroGC wrote on 2023-05-03, 06:32:

I'll attach here all the conversion. I'm always getting volume differences using the same player. All the s98 file I've attacche […]
Show full quote

I'll attach here all the conversion. I'm always getting volume differences using the same player. All the s98 file I've attacched I've created with the new release of your converter.
Inside the Zip, the original VGM files from VGM rips, my edited files with mixed channel (thanks to vgm-edit.py).
Thank you for your help.
Aldo

Yeah, those VGM rips have YM2203 commands (0x55) and YM2608 part 1 commands (0x56) mixed together. I don't know if a PC-98 can have both sound chips installed at the same time or what. But the header doesn't have a YM2608 clock so it doesn't even make sense.

Anyway, I changed the program again so that YM26xx tracks can be split into two separate tracks, using the -one and -two switches. Maybe that will help, since it sounds like your playback software can't play YM26xx tracks at all?

bakemono wrote on 2023-05-03, 12:35:

Yeah, those VGM rips have YM2203 commands (0x55) and YM2608 part 1 commands (0x56) mixed together. I don't know if a PC-98 can have both sound chips installed at the same time or what. But the header doesn't have a YM2608 clock so it doesn't even make sense.

The YM2608 is backward-compatible with the YM2203 (which in turn is backward-compatible with the YM2149, which is effectively a AY-3-8910).
So perhaps it treats the first 3 channels as a regular YM2203, and uses YM2608 commands for the extra channels?

As for headers not containing the clock for chips they use... that may be a separate issue. I've found some older rips for AY-3-8910 that suffered the same issue.
I moved to running a scanner over the entire VGM to see which commands are used, as a failsafe way of determining which chips it uses.

bakemono wrote on 2023-05-03, 12:35:

RetroGC wrote on 2023-05-03, 06:32:

I'll attach here all the conversion. I'm always getting volume differences using the same player. All the s98 file I've attacche […]
Show full quote

I'll attach here all the conversion. I'm always getting volume differences using the same player. All the s98 file I've attacched I've created with the new release of your converter.
Inside the Zip, the original VGM files from VGM rips, my edited files with mixed channel (thanks to vgm-edit.py).
Thank you for your help.
Aldo

Yeah, those VGM rips have YM2203 commands (0x55) and YM2608 part 1 commands (0x56) mixed together. I don't know if a PC-98 can have both sound chips installed at the same time or what. But the header doesn't have a YM2608 clock so it doesn't even make sense.

Anyway, I changed the program again so that YM26xx tracks can be split into two separate tracks, using the -one and -two switches. Maybe that will help, since it sounds like your playback software can't play YM26xx tracks at all?

I use Kumatan MDPlayer for testing, but my real hardware just works with YM2203 s98 files.
Kumatan did for me even a VGM to S98 coverter, but it didn't work because of the S98 version 3 output (while my real hardware just read s98 version1).
I don't think PC98 could have both chip. I'm trying to have an s98 sounding exactly as the VGM (the mixed one), before using it on my hardware for recording the soundtrack.
Aldo

Scali wrote on 2023-05-03, 13:06:

The YM2608 is backward-compatible with the YM2203 (which in turn is backward-compatible with the YM2149, which is effectively a […]
Show full quote

The YM2608 is backward-compatible with the YM2203 (which in turn is backward-compatible with the YM2149, which is effectively a AY-3-8910).
So perhaps it treats the first 3 channels as a regular YM2203, and uses YM2608 commands for the extra channels?

As for headers not containing the clock for chips they use... that may be a separate issue. I've found some older rips for AY-3-8910 that suffered the same issue.
I moved to running a scanner over the entire VGM to see which commands are used, as a failsafe way of determining which chips it uses.

I considered whether it might be a game that supports both sound boards, and maybe has some spurious writes happening that simply do nothing when only the YM2203 is present. That would be believable if the YM2608 writes were going to part 2, but in these files the writes are to part 1 (which is the part that is backwards compatible, the 3 additional FM channels are in part 2).

For anyone still interested in this topic...

People are pushing OPL to new limits with modern software like Furnace Tracker. Check this out:
https://twitter.com/64Abstract/status/1690619630304546816 (although this is not pure FM)

I love this OPL3 cover - https://chiptune.app/?play=OPLArchive%2FA%20M … 27s%20Theme.vgz
it is a cover of Fighting Layer on Namco System 12 - https://vgmrips.net/packs/pack/fighting-layer … 8-janis-luciani

Finally, check out the excellent OPL Archive: https://opl.wafflenet.com/
I have mirrored it on Chip Player as well. https://chiptune.app/browse/OPLArchive