FMK song import accuracy

[GoogleCodeExporter]

http://kulor.arnoldascher.com/TFM/
FMK import is rather broken, portamentos in particular.

Original issue reported on code.google.com by pissmasterplus on 19 Apr 2014 at 9:55

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

I'll have a look what is the issue with tone portamento (at the time of coding, 
there were only few 'demo' songs in FMK format available), but there are more 
'deviations' of FM-Kingtracker to the ordinary processing of effect commands. 
Generally, it's not possible for one tracker to 100% accurately play such 
variety of different file formats from different trackers made by different 
coders. There are some restrictions for those 'outstanding' files which are not 
treating common effect command information as the rest (majority) of trackers, 
incl. AT2. Particulary for FMK, there is following information stated in the 
.doc file for file format conversion:

"After conversion, the Tremolo and Vibrato effects may sound different,
 because FM-Kingtracker uses different speed tables and optional
 waveform definitions that are currently not supported.
 Also the Retrig Note effect is slightly different (if i should be honest,
 i really couldn't get any sense of Sami's frame counting; all i could
 do to make this effect sound way "authentic" was the frame correction during
 conversion phase--and this ain't perfect, though :)
 The OPL3 setting is ignored, Stereo setting is accepted, Rhythm mode is
 not supported, because it was incorrectly implemented in earlier
 versions of the tracker, and the author himself stopped supporting it.
 Also the pattern order list will be truncated to 128 if exceeds."

Original comment by subz3ro....@gmail.com on 20 Apr 2014 at 10:49

[GoogleCodeExporter]

Setting Tone Portamento speed value 3 times of the original seems to be fine 
for my ears (2 was too less, 4 was too much). Please compare yourself and give 
me some feedback. Thanks.
P.S. Forced OPL3 port range 1-FFFFh is also introduced in this test version.

Original comment by subz3ro....@gmail.com on 22 Apr 2014 at 7:54

Attachments:

• ADTRACK2.ZIP

[GoogleCodeExporter]

Welp, I'm gonna have to make some test tunes to properly compare. I can say 
this so far, from Kulor's tunes:

*menuone.fmk
Pat 3; chan 7, 8: 1xx strength too high.
Pat 6, 7, 8; chan 8: echo channel is too loud.

*cyber.fmk
Pat 2 and onwards; chan 3, 4: 2xx strength too low.
Pat A and onwards; chan 12, 13: Many things that don't sound right.

Original comment by pissmasterplus on 22 Apr 2014 at 5:26

[GoogleCodeExporter]

cruisin.fmk/DISKO4.fmk does sound better now, since it uses 3xx command.

Original comment by pissmasterplus on 23 Apr 2014 at 1:01

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

FMK conversion should be fixed. Please verify. Thanks!

Original comment by subz3ro....@gmail.com on 26 May 2014 at 1:22

Attachments:

• ADTRACK2.EXE

[GoogleCodeExporter]

Sounds better now. Although there are still a few things.
CYBER.FMK:
PAT 09 0A
CHN 12 13
Portamento there doesn't sound right.

PAT 06
CHN 04
Loud instrument

SPASE.FMK
PAT 01
CHN 17 18
Wrong arpeggio value

PAT 06 07 08
CHN 16 17
Lead slightly louder. Probably same volume issue.

Original comment by pissmasterplus on 27 May 2014 at 8:52

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

* As CYBER.FMK, PAT 09 0A, CHN 12 13, Portamento sounding not right
It's a bug in FM-Kingtracker, notice garbage effect info there after Frequency 
Slide Down effect. AT2 was interpreting command info to this last command used. 
Such garabage info is ignored now. FIXED.

* As the loud instrument volume issues:
Volume scaling is now turned on for FMK modules. It's nowehere mentioned in the 
tech. doc's that the FM-Kingtracke uses it, but it looks now that it does. 
FIXED.

* As the arpeggio value being wrong -> FIXED

Original comment by subz3ro....@gmail.com on 27 May 2014 at 2:50

Attachments:

• ADTRACK2.EXE

[GoogleCodeExporter]

Great

CYBER.FMK
PAT 0F
CHN 12 13

Silence, lead not playing.

Original comment by pissmasterplus on 27 May 2014 at 3:34

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

Ok. Let's say we will take over this interpretation of Tone Portamento with 
KeyOff-ed channel to AT2. Try again... ;)

Original comment by subz3ro....@gmail.com on 28 May 2014 at 12:33

Attachments:

• ADTRACK2.EXE

[GoogleCodeExporter]

Awesome.

There could be other, much more subtle bugs, but I am getting lazy and want to 
leave for some other time.

Original comment by pissmasterplus on 28 May 2014 at 1:15

[GoogleCodeExporter]

Well, still we have an issue with Vibrato/Tremolo because of very *weird* speed 
table used by FM-Kingtracker. It's very very slow within first 16 values, which 
are listed in the documentation, but it is also stated that the table size is 
256 (!!!)
It's not so far possible to be 100% accurately convert this to AT2 while we are 
using speed tables with size 32 (which was kinda standard for all trackers). 
Maybe it would be possible to emulate with vibrato macro tables, but this is 
not worth of the effort I think.

Original comment by subz3ro....@gmail.com on 28 May 2014 at 1:23

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

So this means not just vib/trem values resulting in different speeds/depths 
between the trackers but also values unreachable by AT2?

Original comment by pissmasterplus on 28 May 2014 at 5:52

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

Ok, I managed to *enhance* Set custom speed table effect so it supports also 
more than 32-byte speed tables. Now the FMK vibrato seems to sound fine with 
128-byte speed table and speed processing factor x3. Please check yourself and 
let me know... :-)

Original comment by subz3ro....@gmail.com on 30 May 2014 at 9:54

Attachments:

• ADTRACK2.EXE

[GoogleCodeExporter]

https://dl.dropboxusercontent.com/u/16644842/AT2/ADT2_FMK_comparisons.rar

Right, I did a few dosbox recordings of AT2 and FMK playing the same thing(in 
that order) and stitched halves together.
It's mostly accurate now, although:
-Probably lots of little off-by-one errors in instrument interpretation.
-I think panning is reversed and not of same strength.
-portamento strength is very close, falls a bit short.
-FMK probably has slightly different implementation of arpeggio effect, as 
evident in MENUONE.

Original comment by pissmasterplus on 30 May 2014 at 12:59

[GoogleCodeExporter]

>panning strength
Oh god what? Silly PissMasterPlus, OPL3 has no fine panning!

Original comment by pissmasterplus on 30 May 2014 at 1:07

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

Panning is completely reversed in FMK (try with stereo.fmk or with using S0x 
effect), so checking 'Flip L/R' option is a must-have for it... then the 
panning will sound correctly :-)
As the errors in instrument interpretation, I don't know what exactly do you 
mean, but I'm not aware of any bug found in instrument interpretation with AT2 
over those years, so probably the issues are on FMK's side :-)
As portamento strength - I think there are some general issues with frame 
counting for effects during playback in FMK. I'm afraid that without having 
source code of replay routine, it's not possible to find out what is the 
difference, and I hardly believe this can be somehow treated during conversion 
to AT2, though. :-(

Original comment by subz3ro....@gmail.com on 30 May 2014 at 6:17

[GoogleCodeExporter]

In SPASE01.WAV, lead echo thing seems to sound slightly thinner in ADT2.

Original comment by pissmasterplus on 30 May 2014 at 9:10

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

Can you make both OPL3 sound captures (FMK and AT2) for comparison? I don't 
have DOS machine anymore and with OPL3 emulation the difference is not 
significant enough. As in, I can hear some difference but don't know if it's 
only different volume or something else. Thanx.

Original comment by subz3ro....@gmail.com on 2 Jun 2014 at 9:30

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

https://dl.dropboxusercontent.com/u/16644842/AT2/SPASE01_short.rar
Recorded from ESSFM. I think I might be better off recording the individual 
instrument.

https://dl.dropboxusercontent.com/u/16644842/AT2/SPASE01_short_lead.rar
There we go.

Also, I misspelled poopeggio.

The weird arp popping in MENUONE is an FMK specific feature and can probably be 
left behind. The arp handling itself does seem different, though.

Original comment by pissmasterplus on 3 Jun 2014 at 6:01