xmp-openmpt: An XMPlay input plugin based on OpenMPT

guest

There is an inaccuracy in the playback of GMC:

amos.gmc
sample 1 from 0:11

miami chase (plant).gmc
sample 7, 10 from 0:08

zd-breathingvectors-2.gmc
sample 2 from 0:02

saga

Oops, loop lengths were treated as bytes instead of words, this should be fixed now.

guest

Silly.med
• instrument 6 isn't played accurately eg. at 0:02, 0:05, 0:08, ...
• sounds on channels 3 and 4 aren't cut off at 1:17
• the title should be "Silly", not "Intro"

Jungle_Ballad.med
• long instrument names aren't displayed in full
• the title should be "Jungle_Ballad"

saga

Quote from: guest• long instrument names aren't displayed in full

Internally, sample or instrument names longer than 31 characters are currently not supported. This may change in the future, but not now.

Quote from: guest• the title should be "Silly", not "Intro"
• the title should be "Jungle_Ballad"

If there's only one song in the file, the "global" song title instead of the (potentially automatically determined) subsong title is now used.

saga

"Silly" should play more accurately now.

guest

MED.Flowing Sky

• Stereo separation cannot be set according to OctaMED Soundstudio.
The stereo separation slider in the plugin seems to have no effect.

• Instrument J (tonfo) eg. in row 6, 38 on 4th channel of first order
and instruments on last channel of eg. 23rd, 27th order
are played too quietly.

saga

Quote from: guest• Stereo separation cannot be set according to OctaMED Soundstudio.
The stereo separation slider in the plugin seems to have no effect.

The stereo separation in the plugin has no effect because there is no stereo separation at work here. The file contains a pan table for the individual tracks, which is all-center. The docs don't mention anything about this, but I guess this pan table needs to be ignored when the "free pan" flag is not set for MMD2 files. In MMD3 files it appears that the panning table must be used regardless of the freepan flag.

Quote from: guest• Instrument J (tonfo) eg. in row 6, 38 on 4th channel of first order
and instruments on last channel of eg. 23rd, 27th order
are played too quietly.

Another one of those undocumented changes between (Octa)MED versions. In MED Soundstudio for Windows, the volume range is 0 to 127 while on Amiga it's 0 to 64. I am not sure if using one or the other interpretation is only up to the format version (MMD2 vs MMD3), so the fix that I applied might be too general.

guest

OctaMED Soundstudio V1.03c seems to not open this MMD2 in Mix mode by default (unless "Use Mixing (MMD)" in Miscellaneous Options is switched on) but stereo separation slider in plugin has no effect.

manx

2024-07-21: xmp-openmpt version 0.7.9 released!

* [**Sec**] Potential division by 0 when seeking in the module with
`seek.sync_samples` enabled (r21167).

* [**Change**] The work-around for
<https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115049> has been changed from
forcing `-O1` on GCC 14 to setting `-fno-ipa-ra` on all GCC versions on
non-ELF platforms. We are still not 100% sure if this work-around is
sufficient in all circumstances. If you are using a non-ELF platform, it is
strongly recommended to update GCC to versions 12.5.0, 13.4.0, 14.2.0, or
15.1.0 as soon as they are released, or to apply the patch from the linked
GCC issues.

* MOD: Allow sample swapping to work when swapping from a non-looping, stopped
sample back to a looping sample (fixes MOD.energy).
* DBM: Import second sustain point in case the first sustain point is not set,
or if it has a lower index than the first.
* DBM: When several instruments referenced the same sample with different
properties (volume, loop points, etc.), only one set of properties was
imported (fixes DBM.Supernova).
* DBM: Prioritize effects more correctly when the same effect is encountered
in both effect columns of a cell (fixes DBM.143_Gnoj).
* DBM: Don't import offset effects when there's a tone portmento next to them.
* DBM: A few IT-specific playback quirks are disabled for more accurate
playback (e.g. in "Are You Flying With Me?" by Jazzcat).
* DIGI: Sample play direction was reset if adjacent channel contained a
Note Cut note.
* AMF: When running out of sample slots, file reading became be misaligned
because the sample name was not skipped.
* MED: Command 0F was not imported.
* MED: Upper frequency limits should be more accurate now.
* MED: Channel panning is now only applied in MMD2 files if the free pan flag
is set.
* MED: Volume command resolution was incorrect for pre-MMD3 files.
* XM: oggmod does not support stereo samples but keeps the stereo flag when
encoding such samples. Such samples are now imported as mono samples instead
of not importing them at all.
* XM: For files saved with registered MadTracker 2 versions, do not put
binary garbage (the user ID) in the tracker metadata field. It is replaced
with "registered" instead.
* For some truncated files, the used tracker was not identified correctly.
* S3M: Identify files saved with early Impulse Tracker versions, Sound Club 2,
Liquid Tracker, NESMusa, UNMO3, deMODifier, Kosmic To-S3M, and better tell
old ModPlug Tracker versions apart.
* S3M: When skipping sample loading, some tracker identifications were not
working as intended.
* IT: Identify files saved with itwriter.
* DTM: Identify files saved with Digital Tracker 2.3.

* xmp-openmpt: If there is only one subsong, set the song title to the
"global" song title instead of the name of that subsong.
* xmp-openmpt: Sample and instrument names were not sanitized, sometimes
showing on multiple rows.

See https://lib.openmpt.org/libopenmpt/2024/07/21/security-updates-0.7.9-0.6.18-0.5.32-0.4.44/

Downloads:
* xmp-openmpt for Windows 10 21H2 (or later) and SSE2-capable CPU, or legacy version for Windows 7 SP1 (or later) and SSE2-capable CPU
* xmp-openmpt RETRO for Windows XP SP3 (or later) or non-SSE2-capable CPU
* xmp-openmpt RETRO for Windows 98 SE + KernelEx (or later) (without configuration dialog)

saga

Quote from: guestOctaMED Soundstudio V1.03c seems to not open this MMD2 in Mix mode by default (unless "Use Mixing (MMD)" in Miscellaneous Options is switched on) but stereo separation slider in plugin has no effect.

Again, the stereo separation has no effect because it it is a user setting for the widening or narrowing the final mix. That is, if a channel panning table with all-center panning was loaded, stereo separation cannot undo that. I'm still not sure which Amiga versions of OctaMED or MED Soundstudio actually support this panning table, because in MED Soundstudio I cannot find it, despite version 1.03c being newer than the document that describes the free panning table. So for the time being I'll have to assume that it should only be used if free panning is enabled and hardware mixing is disabled. Let's see what breaks next...

guest

• MED.(Liberation)intro
‣ Instrument 1 e.g. in the first channel of the first order should not be muted after several rows
‣ Plugin reports duration of 5:26 but stops playing at 7:18

• Plugin reports duration of 1:55 for the second subsong but stops playing at 2:56

• Plugin reports duration of 3:51 but OctaMED Soundstudio stops playing at 4:05 (see e.g. 3rd order from row 67)

BTW, could the plugin display the current order number on the Pattern Display or in main window like XMPlay does?

saga

Hybrid instrument loops should now be imported correctly, and length calculation inconsistencies in MED (and also MDL) should be gone too.

saga

Song length when different subsongs have different initial speed or tempo is fixed (all formats), as well as MED command 09 with parameters > 20 and <= 32.

guest

ok.sfx
⁍ channel 4 at 3rd position isn't played accurately

daylight.sfx ⁍ clicks on channel 3 at first position
freeflight.sfx ⁍ clicks on channel 4 at first position
freedom91.sfx ⁍ clicks on channel 2 at second position

saga

SFX arpeggios are now emulated more accurately, and oneshot samples in SFX files are now shortened if the sample header indicates a smaller size than the actual sample data size.

guest

MED.Untitled6
‣ vibrato doesn't sound quite accurate eg. on two last channels at 4th position

MED.Aria
‣ plugin doesn't play subsongs 6-19, 21

saga

Both problems should be fixed.

guest

· It seems that the tempo at the beginning of the 3rd position in this MED is set incorrectly.

· Portamento up reaches too high on channel 7, 8 at first order in this 669.

· The title of this 669 isn't displayed in full.

saga

The first two issues are now fixed, the third one will require a bit more work for a proper fix and an issue was created to track that: https://bugs.openmpt.org/view.php?id=1813

guest

Quote from: guest· It seems that the tempo at the beginning of the 3rd position in this MED is set incorrectly.
Quote from: saga... fixed

• Plugin reports duration of 3:53 but but the playback stops at 1:23.

• on our way.669
Sample 5 ← channel 3, 4 ← order 0 is played too high.

• Plugin plays this OKT too slowly at the beginning.

• bugfixed (the song).ptm
Starting at 2:26, PolyPlayer plays bass deeper.

saga

Quote from: guest• bugfixed (the song).ptm
Starting at 2:26, PolyPlayer plays bass deeper.

It sounds identical to me, compared to both PolyPlayer and PolyTracker? Can you provide a recording of what you hear in xmp-openmpt and what you hear in PolyPlayer?

Edit: I think I figured it out. There's a bug in PT's panning command where extreme panning positions cause the note to played at higher volume.

QuotePlugin reports duration of 3:53 but but the playback stops at 1:23.

The correct length is now reported.

Quote• on our way.669
Sample 5 ← channel 3, 4 ← order 0 is played too high.

Effects were quite messed up after the last 669 fix, now they should be correct again.

Quote• Plugin plays this OKT too slowly at the beginning.

This should also play as intended now.

guest

Quote from: guest• bugfixed (the song).ptm
Starting at 2:26, PolyPlayer plays bass deeper.

After recent fixes, changing position with slider doesn't work accurately (eg.
bass before starting at 2:26 starts at ~2:54 after changing position with slider).

saga

This broke due to the MED song length calculation fix, and should now hopefully work again.

guest

So-So-Mahhn-REmix.med
Sample 13 ← 3rd channel ← second order doesn't seem to be played accurately.

saga

It seems like MED Soundstudio for Amiga and Windows disagree on how to play that sample. OpenMPT plays it the way the Windows version does.

Edit: And without knowing which of the two programs was used, the Windows version definitely sounds more like it was intended to sound like that.