Messages

IT'S READY!

I uploaded the S3M conversions to the AMP FTP, under the filename 'Jason_Page.zip'.

In the archive are the soundtracks to Empire Soccer 94, Fire and Ice (incl. Xmas special), Nipper vs. The Kats, Uridium 2 and Virocop AGA.

Note: According to Hall of Light, the Virocop music was composed by Lee Banyard.

Jason Page is one of those formats where every file is divided into 'subsongs', and my converter makes each one of those into a separate file, so in some cases there are many .s3m files for each .jpn one.  The subsong is indicated by a suffixed hyphen and the number.

Be aware that a few of the files (fire and ice-7-1 for example) exceed the ScreamTracker 64,000 byte sample limit.  This shouldn't cause problems for well-written software (my player had no trouble), but watch out all the same.

Off to write loads of documentation now...

You mean you'd do that?  Cool!  I was worried I'd have to beg.

Strictly speaking, they're not 'true' conversions, because the Jason Page format is so unusual, but AFAICT they sound more or less exactly the same and they're a whole lot more compatible.  The main disadvantage is that the sound quality is only possible by a rather large (100s of KBs) file size.

<Farnsworth voice>
Good news, everyone!

After no small amount of help from the Amiga Hardware Reference Manual, the source code from this website and the Motorola 68000 ARM, the core part of the converter is finally ready.

All that remains to be done is convert it to output ScreamTracker files, and, of course, write the documentation.  (I still haven't found a good place to upload my documentation for the Onyx format - according to Google the only mentions of it on the web are this forum and another one which the OP of that thread tried first.)  After that I'll email you the conversions (or would it be better to use FTP?).  I assume Asle didn't have any luck with the Digital Symphony files, but I have now finally managed to obtain the ScreamTracker documentation, so I should be able to handle the conversion myself.

Next up I intend to have a good crack at the Richard Joseph format.  (I remember Diggers in particular with some fondness.)

Have fun and let us know if you have any further question.  

I have a further question.

This really needs to be looked at by someone who has previously programmed in 68000 assembly language - I'm learning it by staring hard at it, which has its disadvantages!

I can't work out this code:

Code Select Expand


  MOVE.L (A3)+,D0
  MOVE.L D0,(_SampleLoopLen,A2)
  BEQ.B lbC000C3A
  LSR.W #1,D0
  MOVE.W D0,(_SampleLength,A2)
lbC000C3A
  RTS


If I'm reading it right (which I'm probably not), this translates to: Load 4 bytes from address A3, store them in address (A2 + _SampleLoopLen), then, only if the 4 bytes are not 0, divide them by 2 and store them in address (A2 + _SampleLength).

There is another function which does the same, but doesn't check if the bytes are 0 first.

This therefore means that the only value address (A2 + _SampleLoopLen) can take is 0, or (A2 + _SampleLength) * 2.

Which makes the code to program the Paula chip rather puzzling:

Code Select Expand


  TST.L (_SampleLoopLen,A2)
  BEQ.B lbC000A98   ; silence channel
  MOVEQ #0,D0
  MOVE.W (_SampleLength,A2),D0
  ADD.L D0,D0
  MOVE.L (_SampleLoopAddress,A2),D1
  ADD.L D0,D1
  MOVE.L (_SampleLoopLen,A2),D0
  SUB.L D0,D1
  MOVE.L D1,(_HSampleStartAddr,A4)
  MOVE.W (_SampleLength,A2),(_HSampleLength,A4)

(Snipped for relevance and brevity.)

...as this means the address of the sample data can only take the value of (A2 + _SampleLoopAddress) (since the function is bypassed if _SampleLoopLen = 0).

...which makes you wonder what the point of _SampleLoopLen is.

I really hope I'm wrong.

Most wonderful news indeed!  I'll get to work on it now.

Excellent!  I'll email them now.

You can contact me at:

martin [dot] bazley [at] blueyonder [dot] co [dot] uk

Thanks a lot for that!

When you say "send the tunes over to Asle", do you mean the original format tunes, or the Digital Symphony conversions?  (The original format tunes are all available on Modland - that's where I got them from.)

Hmm, apparently the answer is no.

I have now managed to produce a set of files which are a fairly close approximation to the sound produced by the original player.  Finding the source code for the Wanted Team's conversion was definitely a big help.

The only problem is that, due to the way the converter works in order to give the best sound possible, several of the files exceed the ProTracker 31 sample limit.  Although it doesn't use many effects, as most of the Jason Page effects are mutually incompatible with the MOD format and its derivatives (I ended up encoding them in copies of the sample data instead), it really does need a format with a long sample list.  At the moment I'm using Archimedes Digital Symphony, and as the only non-RISC OS program I know of which can play that is xmp, which is a command-line program, I'd like to be able to convert them to another format.

The one which seems most sensible is ScreamTracker 3 (.S3M), as that has up to 99 samples, but I don't have any tools to make those files.  Can anybody help?  I've attached a file as an example, so if anybody could return this converted to S3M (or another format with similar features), that'd be great and I'll upload the rest to Modland.

Alternatively, if somebody could confirm it's OK to send files to AMP in such an obscure format, that'd be great as well.

Thanks.

Many congratulations, and here's to the next generation of hopeless computer geeks!

Reading Dr. Awesome's interview reminded me - would it be possible to add an extra field to the composers database for personal websites?  Also, I see deus-ex provides links in the data posted to the 'musicians' forum...

Hi,

I've been spending a lot of my spare time recently decoding the so-called 'Jason Page' tracker format used in Fire and Ice.  Unfortunately, I only had access to the Archimedes conversion of the game's internal playroutine, in ARM assembler, and as I got deeper in it became more and more apparent that the person who did the conversion didn't know the first thing about what he was doing or the relative properties of PC, Amiga and Acorn sound architecture.

So, while I do now have a converter program that produces music and a fairly thorough understanding of how the format is structured and how it is read, I am aware that the results, while faithful to the Acorn version, are completely different from the original one.

I'm asking because I'd like to know if anyone has any information on this format (or, indeed, a way to ask Jason himself about it) so that I could make my conversion a good deal more faithful.  I couldn't find anything on Exotica (there's a page for it, but it's blank), but I hope I'm not the only person who knows anything about this format - the only alternative would be to get a look at the Amiga binary, and I can only read ARM assembler!

Argh ... Xlat just showed me there's another pb.
It's about the pattern break command (D**). It must contain decimal figures, while, in here, it's hexadecimal. Shows pretty well in the hambug-tra music. *geee* ... I'll update the source (again) and list the affected musics.

Whoopsy!  Good job somebody pointed that out - although I was listening to Humbug-tra earlier today, and I did wonder...

Slightly modified format specification attached.

Except for your conversion of shiva_furnace.mod which sounds badly wrong (ouch, my ears ), and the rhythm of the beat that starts at order 01 is off. You certainly intent to punish us, do you?

Yep - I recognise that problem.  All the finetunes have been blanked.  'Battle squadron' has a similar problem.  Taking another look at the C code, I can't see anything to write the finetune values anywhere.  It's a signed 16-bit little endian value at offset 22 in the sample info block, and it's the MOD range (-8 to 7) multiplied by minus 487.  Took me some time to work it out!

@deus-ex: Incidentally, how did loading the file into Milkytracker and saving it to ProTracker set all the finetunes?  That must be one powerful program!

EDIT: Dammit!  Little endian, sorry.  I'm just not used to it being the exception, rather than the rule...

Great job - thanks!  (I knew my conversion wasn't quite accurate, but it sounded like it should have done on anything I played it with... Bearing in mind the problems with more than 4 channels, TakeTracker is probably the best option.)

(EDIT: Has anyone tried to contact @larryhagman?  He was the one who wanted this in the first place, after all...)

OK.  I have now uploaded (in the '8ChanExp' directory) the defective modules in 'incorrect' format to MODLAND, just to make sure.

Sorry, next reply will be a while - my parents are insisting that I eat something!