Author Topic: xmp-openmpt: An XMPlay input plugin based on OpenMPT (Read 168058 times)

guest · « **Reply #325 on:** 5 Nov '24 - 17:32 »

Plugin does not play sample 20 starting from 0:06.

saga · « **Reply #326 on:** 5 Nov '24 - 17:42 »

This was caused by the Amiga resampler being used for MED files by default, and this sample (plus a few others in that file) contains stereo samples with identical but phase-flipped data on the two sample channels. Since the Amiga resampler does not support stereo samples, they are downmixed to mono, which causes the left and right channel to be added together and effectively cancel each other out. The Amiga resampler is now disabled for MED files with stereo samples.

guest · « **Reply #327 on:** 6 Nov '24 - 16:36 »

From the very beginning, the plugin does not play a specific instrument.

saga · « **Reply #328 on:** 6 Nov '24 - 21:20 »

This should now be playing correctly. I also found that in the aftermath of the previous FC/MED fix, instruments were now completely detuned in XM files with linear slides and a few other formats; this is now fixed as well.

guest · « **Reply #329 on:** 7 Nov '24 - 20:37 »

This MED has a duration of 5:45 in OctaMED Soundstudio.

saga · « **Reply #330 on:** 7 Nov '24 - 22:49 »

Oh joy, another one of those undocumented behaviours that even differ between OctaMED versions (the Windows version plays the file too slow). I have added a workaround for software-mixed files in SPD mode trying to use a tempo value less than 10, but no idea if once again it's too general or too specific.

guest · « **Reply #331 on:** 12 Nov '24 - 18:08 »

First order sounds slightly different in stereo than in OctaMED.

saga · « **Reply #332 on:** 16 Nov '24 - 00:40 »

Seems like there is yet another opaque condition when volume command parameters are 7-bit vs 6-bit... This is fixed now based on some behaviour changes I found in MED Soundstudio for Windows, as always there is no way to be sure if this is correct though or if it breaks other modules.

guest · « **Reply #333 on:** 17 Nov '24 - 11:58 »

• Looped sample 6 on channel 4 plays differently in Scream Tracker 2.3

• Compared to OctaMED v4:
∙ Sample 5 should not be muted at 0:15 in second subsong and at 0:13 in first subsong.
∙ Sample 6 is played differently from 0:57 in first subsong.
∙ Sample 4 is played differently from 0:00 in first subsong.

saga · « **Reply #334 on:** 17 Nov '24 - 22:28 »

Quote from: guest on 17 Nov '24 - 11:58

• Looped sample 6 on channel 4 plays differently in Scream Tracker 2.3

Apparently ST2 does not sanitize the loop end, so the sample loop continues playing into the next sample. This is now emulated.

saga · « **Reply #335 on:** 18 Nov '24 - 18:06 »

Quote from: guest on 17 Nov '24 - 11:58

• Compared to OctaMED v4:
∙ Sample 5 should not be muted at 0:15 in second subsong and at 0:13 in first subsong.
∙ Sample 6 is played differently from 0:57 in first subsong.
∙ Sample 4 is played differently from 0:00 in first subsong.

These should all be fixed as well.

guest · « **Reply #336 on:** 19 Nov '24 - 19:13 »

‣ Instrument 15 is muted too early in two places in the first order
‣ Instrument 27 is played differently from 3:56 (25th order)

BTW, it would be nice to have assignable keyboard shortcuts for increase/decrease stereo separation.

saga · « **Reply #337 on:** 19 Nov '24 - 21:29 »

Okay, here's another attempt at guessing when to enable 7-bit volumes and when not to. This should fix instrument 15. Instrument 27 is not fixable with OpenMPT's current pattern effect architecture.

saga · « **Reply #338 on:** 20 Nov '24 - 20:46 »

I found a way to slightly improve playback of instrument 27, it's not accurate but better than before.

guest · « **Reply #339 on:** 23 Nov '24 - 13:00 »

There are clicks eg. in the first position.

saga · « **Reply #340 on:** 23 Nov '24 - 14:38 »

If you turn off volume ramping, some clicks can be heard because there is a bass sample that is retriggered constantly and each time it retriggers, it is at a very high point. Volume ramping is exactly meant to reduce clicks like these.

guest · « **Reply #341 on:** 24 Nov '24 - 11:27 »

The beginning of 4th channel in 4th position seems to have no sound in OctaMED.

saga · « **Reply #342 on:** 24 Nov '24 - 12:42 »

Portamento behaviour should be more in line with OctaMED now.

guest · « **Reply #343 on:** 25 Nov '24 - 21:44 »

First channel in 10th order starting from row 28 is played differently in OctaMED.

guest · « **Reply #344 on:** 27 Nov '24 - 16:48 »

OctaMED does not slow down playback so much at the end of the 12th position.

manx · « **Reply #345 on:** 1 Dec '24 - 15:31 »

2024-12-01: xmp-openmpt version 0.7.12 released!

* [**Bug**] Fixed various undefined behaviour found with ubsan.

* IT: Don't report files claiming to be made with Impulse Tracker 2.08+ as IT-made if they have no edit timer.
* IT: Ignore sample data in slots that don't have the "sample data present" flag set, if the file vaguely looks IT-made to avoid playing garbage caused by an Impulse Tracker bug that should not be audible.
* S3M: Detect early Schism Tracker versions.
* MOD: When trying to detect MOD files with broken order lists, the file size is now rounded down to an even number. This helps identifying some malformed files (MOD files can technically not have an odd size).
* MOD: Also enable ProTracker-compatible tremolo ramp waveform for M!K! modules.
* MOD: In ProTracker 1/2 mode, retrigger with instrument-less notes now keeps using the previous sample offset.
* Warn when a Startrekker AM file most likely requires an (currently unsupported) external instrument definition file.
* DBM / IMF / MED: When merging pattern commands, allow to move offset to volume column at the expense of a lower offset resolution.
* MED: Fix correct octave transposition in some MED files that have hardware mixing disabled but sample transpose enabled.
* MED: Don't enable Amiga resampler if any stereo samples are found, as it does not support stereo samples.
* MED: Fix tempo in some files using software mixing mode and legacy tempo values.
* MED: Avoid importing effect memory for some commands.
* MED: Retrigger with instrument-less note now keeps using the previous sample offset.
* MED: Disable sample swapping on notes with portamento, and don't resume stopped notes with portamento.
* MED: Only use 7-bit volume commands in MMD3 files made with a new enough version of MED Soundstudio.
* STM: Do not sanitize sample loop data. Scream Tracker 2 reads into the next sample's data when loops exceed the sample length.
* When evaluating MIDI macros containing letters "u" or "v" during seeking, the initial global volume was applied to them, rather than the global volume that would be reached at that pattern position.

* mpg123: Update to v1.32.9 (2024-11-02).

See https://lib.openmpt.org/libopenmpt/2024/12/01/releases-0.7.12-0.6.21-0.5.35-0.4.47/

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)