- Update to version 0.4.0:
* [New] libopenmpt now includes emulation of the OPL chip and
thus plays OPL instruments in S3M, C67 and MPTM files.
OPL chip emulation volume can be changed with the new ctl
render.opl.volume_factor.
* [New] libopenmpt now supports CDFM / Composer 670 module files.
* [New] libopenmpt: New ctl play.at_end can be used to change
what happens when the song end is reached:
- “fadeout”: Fades the module out for a short while.
Subsequent reads after the fadeout will return 0 rendered
frames. This is the default and identical to the behaviour
in previous libopenmpt versions.
- “continue”: Returns 0 rendered frames when the song end is
reached. Subsequent reads will continue playing from the song
start or loop start. This can be used for custom loop logic,
such as loop auto-detection and longer fadeouts.
- “stop”: Returns 0 rendered frames when the song end is
reached. Subsequent reads will return 0 rendered frames.
* [New] Add new metadata fields "originaltype" and
"originaltype_long" which allow more clearly reflecting what
is going on with converted formats like MO3 and GDM.
* [Change] openmpt123: openmpt123 now rejects --output-type in
--ui and --batch modes and also rejects --output in --render
mode. These combinations of options really made no sense and
were rather confusing.
* [Regression] Support for libmpg123 older than 1.14.0 has been
removed.
* [Regression] libmodplug: Support for emulating libmodplug
0.8.7 API/ABI has been removed.
* [Bug] xmp-openmpt: Sample rate and number of output channels
were not applied correctly when using per-file settings.
* [Bug] Internal mixer state was not initialized properly when
initially rendering in 44100kHz stereo format.
* [Bug] openmpt123: Prevent libsdl2 and libsdl from being enabled
at the same time because they conflict with each other.
* [Bug] libmodplug: Setting SNDMIX_NORESAMPLING in the C++ API
always resulted in linear interpolation instead of nearest
neighbour
* IT: In Compatible Gxx mode, allow sample changes next to a tone
portamento effect if a previous sample has already stopped
playing.
* IT: Fix broken volume envelopes with negative values as found
in breakdwn.it by Elysis.
* MOD: Slides and delayed notes are executed on every repetition
of a row with row delay (fixes “ode to protracker”).
* XM: If the sustain point of the panning envelope is reached
before key-off, it is never released.
* XM: Do not default recall volume / panning for delayed
instrument-less notes
* XM :E60 loop bug was not considered in song length calucation.
* S3M: Notes without instrument number use previous note’s sample
offset.
* Tighten M15 and MOD file rejection heuristics.
* J2B: Ignore frequency limits from file header.
Fixes Medivo.j2b, broken since libopenmpt-0.2.6401-beta17.
* STM: More accurate tempo calculation.
* STM: Better support for early format revisions (no such files
have been found in the wild, though).
* STM: Last character of sample name was missing.
* SFX: Work around bad conversions of the “Operation Stealth”
soundtrack by turning pattern breaks into note stops.
* IMF: Filter cutoff was upside down and the cutoff range was
too small.
* ParamEq plugin center frequency was not limited correctly.
* Keep track of active SFx macro during seeking.
* The “note cut” duplicate note action did not volume-ramp the
previously playing sample.
* A song starting with non-existing patterns could not be played.
* DSM: Support restart position and 16-bit samples.
* DTM: Import global volume.
* MOD: Support notes in octave 2, like in FastTracker 2
(fixes DOPE.MOD).
* Do not apply Amiga playback heuristics to MOD files that have
clearly been written with a PC tracker.
* MPTM: More logical release node behaviour.
* Subsong search is now less thorough. It could previously find
many subsongs that are technically correct (unplayed rows at
the beginning of patterns that have been jumped over due to
pattern breaks), but so far no real-world module that would
require such a thorough subsong detection was found. The old
mechanism caused way more false positives than intended with
real-world modules, though.
* Restrict the unpacked size of compressed DMF, IT, MDL and MO3
samples to avoid huge allocations with malformed small files.