-
Notifications
You must be signed in to change notification settings - Fork 383
Guide:Setting up MIDI in DOSBox‐X
Back to the DOSBox-X Wiki Welcome page.
DOSBox-X has support for emulating either a General MIDI or a Roland MT-32 synthesizer. In addition, it is possible to use an external MIDI synthesizer, either real or software.
Note
|
As of DOSBox-X version 0.83.6, the DOSMID program is included in DOSBox-X and will appear on the virtual Z: drive, which can be used to playback MIDI/RMI/MUS audio files. |
Note
|
You can check the currently active MIDI setup in DOSBox-X by going to the Sound option on the menu bar (when in windowed mode), and selecting "Show MIDI device configuration". |
This section will list the available MIDI configuration options that DOSBox-X provides, with exception to the MT-32 and FluidSynth options which are listed in their respective sections.
-
Default value: intelligent
-
Valid values: intelligent, uart, none
The MPU-401 is a PC to MIDI interface, originally developed by Roland.
The default "intelligent" setting, also referred to as "normal mode" Roland MPU-401 emulation should provide the broadest range of compatibility with games.
Setting it to "uart", emulates the later cost-reduced "MPU-401 compatible" options, such as was integrated in various sound cards.
Setting it to "none" disables MPU-401 emulation, and therefore MIDI emulation will be disabled.
Both MT-32 and General MIDI require MPU-401 emulation.
-
Default value: 0
-
Other values - IBM PC: 300, 310, 320, 330, 332, 334, 336, 340, 360
-
Other values - NEC PC-98: c0d0, c8d0, d0d0, d8d0, e0d0, e8d0, f0d0, f8d0, 80d2, 80d4, 80d6, 80d8
This is the base IO address (in hex) of the emulated MPU-401.
If set to 0, the default I/O address for the IBM PC or NEC PC-98 will be used. For the IBM PC this will be 330, while for the NEC PC-98 it will be E0D0.
-
300h to 330h are for use with IBM PC mode.
-
C0D0h to F8D0h (in steps of 800h) are for use with NEC PC-98 mode (MPU98).
-
80D2h through 80DEh are for use with NEC PC-98 Sound Blaster 16 MPU-401 emulation.
-
Default value: -1
-
Other values: 2 - 15
This is the IRQ of the emulated MPU-401 (intelligent mode only).
If set to -1, the default IRQ address for the IBM PC or NEC PC-98 will be used.
-
For the IBM PC XT with a single PIC, this will be IRQ 2
-
For the IBM PC AT (or later) this will be IRQ 9
-
For the NEC PC-98 it will be IRQ 6.
The original Roland interfaces for the IBM PC used IRQ2, and as such many software titles default to IRQ2.
If set to IRQ2 when emulating an IBM AT (286) or better, it will actually be set to IRQ9. However, the IBM PC BIOS redirects IRQ2 to IRQ9 for backwards compatibility, such that software configured to use IRQ2 for the MPU401 IRQ will work.
-
Default value: default
-
Linux Host: "default" will result in it trying to find a MIDI device for output, it will attempt (in order): TiMidity++, FluidSynth and MT-32 emulation.
-
macOS Host: "default" is the same as "coreaudio"
-
Windows Host: "default" is the same as "win32", which in turn will default to the "Microsoft GS Wavetable software synthesizer"
-
Other options:
-
mididevice = win32
-
Defaults to the "Microsoft GS Wavetable software synthesizer" integrated in Windows.
-
Don’t be confused by the name, it also applies to 64-Bit Windows.
-
-
mididevice = alsa
-
Advanced Linux Sound Architecture
-
This option can be used if you have a soundcard with a wavetable, or want to use an external MIDI device.
-
The output device needs to be specified with the midiconfig option.
-
-
mididevice = oss
-
Linux/Unix Open Sound System (deprecated)
-
This option can be used if you have a soundcard with a wavetable, or want to use an external MIDI device.
-
The output device needs to be specified with the midiconfig option.
-
-
mididevice = coreaudio
-
macOS CoreAudio
-
framework to render the music through the built-in OS X synthesizer.
-
A SF2 or SF3 soundfount can be specified with the
midiconfig
setting.
-
-
mididevice = coremidi
-
macOS CoreMidi
-
framework to route MIDI commands to any device that has been configured in Audio MIDI Setup.
-
-
mididevice = mt32
-
Roland MT-32 emulation. See the Roland MT-32 Emulation section below for more details.
-
-
mididevice = synth
-
Deprecated option, same as "fluidsynth"
-
-
mididevice = fluidsynth
-
FluidSynth General MIDI emulation. See the FluidSynth General MIDI emulation section below for more details.
-
-
mididevice = timidity
-
TiMidity++ General MIDI emulation See the TiMidity++ General MIDI emulation section for more details.
-
-
mididevice = none
-
Disable MIDI
-
-
Default value: <blank>
This can be used to pass special options needed for the chosen "mididevice".
For instance, for the mididevice=coreaudio
or synth
options, this can be a path and filename of a SoundFont (in sf2 or sf3 format).
For Linux with mididevice=alsa
or oss
this is the ID or part of the name of the wavetable synth module of your sound card (most sound cards don’t have one), or that of your external MIDI synth.
You can use aconnect -i
to list available devices with ALSA.
You can use the Client ID with device ID combined and set it for instance as midiconfig=14:0
For Windows with mididevice=win32
, this can be used to specify the ID or part of the name of the synth you want to use.
It defaults to the "Microsoft GS Wavetable Synth", which is the software synth integrated in Windows.
In DOSBox-X run mixer /listmidi
to list available options.
For macOS with mididevice=coremidi
, this can be used to specify the ID or part of the name of the synth you want to use (how to list the options?).
In case of a real Roland MT-32 rev. 0 as MIDI output device, some games may require a delay in order to prevent 'buffer overflow' issues.
In that case, add 'delaysysex', for example: midiconfig=2 delaysysex
would cause Windows to use MIDI device 2 (as listed in DOSBox-X with mixer /listmidi
), and apply the delaysysex workaround.
-
Default value: 44100
-
Possible values: 44100, 48000, 32000, 22050, 16000, 11025, 8000, 49716
Samplerate is a deprecated option for use with mididevice=synth
.
Please use mididevice=fluidsynth
instead, with the fluid.samplerate
option if necessary.
You should set this to the same sample rate as the rate=
option in the [mixer]
section.
The Roland MT-32 pre-dates General MIDI, and was used by a broad range of DOS games, starting in 1988 with King’s Quest IV.
The MT-32 emulation that is integrated into DOSBox-X is based on the Munt project.
Emulation is provided for the original MT-32 (aka "MT-32 Old"), the revised MT-32 (aka "MT-32 New"), and the CM-32L and LAPC-I.
Note
|
If a game offers a "Sound Canvas", Roland SC-55, Roland SCC-I or a Roland RAP-10 option you want to try, look at the General MIDI Emulation section below instead. |
There are slight differences between the MT-32 Old and New revisions which in some cases can cause games composed on one to not sound quite right if played back on a different revision. Unfortunately the games themselves don’t specify which revision you need, as the game developers themselves were often not aware of the existence of different revisions.
The CM-32L, from an emulation perspective is an "MT-32 New" with additional sound effects. Games composed on a CM-32L will work on an MT-32, but some sounds may be missing.
The LAPC-I (aka LAPC1), is basically a CM-32L + MPU-401 on a ISA card, and therefore the CM-32L and LAPC-I are identical from an emulation perspective.
To emulate a Roland MT-32 or CM-32L, the original ROMs are needed. Extracting ROM’s from a real Roland MT-32 or CM-32L is documented on the Munt website. This guide will not go into more detail on getting these ROMs.
Ideally you will have two sets of ROMs. One set from the MT-32 Old, and one set from a CM-32L. Since a CM-32L is backwards compatible with an MT-32 New, you don’t really need a ROM set for the MT-32 New if you already have one for the CM-32L.
Once you have the ROMs, save them in different directories. If you save them in the same directory, DOSBox-X will always use the newer CM-32L ROMs, which in some cases will not sound right if the game was composed with an MT-32 Old.
E.g. on Linux
/home/myuser/emu/mt32/MT32_CONTROL.ROM /home/myuser/emu/mt32/MT32_PCM.ROM /home/myuser/emu/cm32l/CM32L_CONTROL.ROM /home/myuser/emu/cm32l/CM32L_PCM.ROM
Or on Windows
C:\Users\My User\emu\mt32\MT32_CONTROL.ROM C:\Users\My User\emu\mt32\MT32_PCM.ROM C:\Users\My User\emu\cm32l\CM32L_CONTROL.ROM C:\Users\My User\emu\cm32l\CM32L_PCM.ROM
Adjust the paths as needed.
At a minimum you need to have the following set in your DOSBox-X config file:
[midi]
mididevice=mt32
mt32.romdir="C:\Users\My User\emu\cm32l"
Adjust the path as needed. In the example above, it points to the CM-32L ROMs, which will cause DOSBox-X to automatically emulate a CM-32L, which will work fine for the majority of games.
You can check the MT-32 compatible games list on Wikipedia for known compatibility issues for games requiring the MT-32 Old ROMs, or games that are better played with General MIDI.
If a PC game asks for the base IO address and IRQ, you should be able to use the default IO 330 and IRQ 9 (or IRQ2 will also work).
Default: <working directory>
This should point to the directory with MT-32 or CM-32L Control and PCM ROM files. Emulation will not work without them.
If no path is specified, DOSBox-X will check the directory from which it was started.
Accepted file names are as follows:
-
CM32L_CONTROL.ROM
andCM32L_PCM.ROM
or
-
MT32_CONTROL.ROM
andMT32_PCM.ROM
If the directory contains both CM-32L and MT-32 ROMs, DOSBox-X will default to emulating a Roland CM-32L.
-
Default: false
-
Possible values: true, false
Reverse stereo channels for MT-32 output
-
Default: 16
-
Valid range: 2-100
Minimum milliseconds of data to render at once (min 2, max 100)
Increasing this value reduces rendering overhead which may improve performance but also increases audio lag.
Valid for rendering in separate thread only.
-
Default: 32
-
Valid options: 3, 4, 32, 199, 200
How many milliseconds of data to render ahead. Increasing this value may help to avoid under-runs but also increases audio lag. Cannot be set less than or equal to mt32.chunk value.
Valid for rendering in separate thread only.
-
Default: 32
-
Valid options: 8, 9, 32, 255, 256
The maximum number of partials playing simultaneously.
-
Default: auto
-
Possible values: 0, 1, 2, 3, auto
MT-32 DAC (Digital to Analogue Converter) input emulation mode. 'auto' equates to '0'.
-
Nice = 0 - default
-
Produce samples at double the volume, without tricks. Higher quality than the real devices
-
-
Pure = 1
-
Produce samples that exactly match the bits output from the emulated LA32. Nicer overdrive characteristics than the DAC hacks (it simply clips samples within range) Much less likely to overdrive than any other mode. Half the volume of the other modes, meaning its volume relative to the reverb output when mixed together directly will sound wrong. So, reverb level must be lowered. Perfect for developers while debugging :)
-
-
GENERATION1 = 2
-
Re-orders the LA32 output bits as in early generation MT-32s (according to Wikipedia). The DAC bit order (where each number represents the original LA32 output bit number, and XX means the bit is always low): 15 13 12 11 10 09 08 07 06 05 04 03 02 01 00 XX
-
-
GENERATION2 = 3
-
Re-orders the LA32 output bits as in later generations (personally confirmed on my CM-32L - KG). The DAC bit order (where each number represents the original LA32 output bit number): 15 13 12 11 10 09 08 07 06 05 04 03 02 01 00 14
-
-
Default: 2
-
Valid range: 0-3
MT-32 analogue output emulation mode
-
Digital = 0
-
Only the digital path is emulated. The output samples correspond to the digital output signal appeared at the DAC entrance. Fastest mode.
-
-
Coarse = 1
-
Coarse emulation of LPF (Low Pass Filter) circuit. High frequencies are boosted, sample rate remains unchanged. Slightly better sounding but also a slightly slower.
-
-
Accurate = 2 (default)
-
Finer emulation of LPF circuit. Output signal is up-sampled to 48 kHz to allow emulation of audible mirror spectra above 16 kHz, which is passed through the LPF circuit without significant attenuation. Sounding is closer to the analogue output from real hardware but also slower than the modes 0 and 1.
-
-
Oversampled = 3
-
Same as the default mode 2 but the output signal is 2x over-sampled, i.e. the output sample rate is 96 kHz. Even slower than all the other modes but better retains the highest frequencies while further resampled in DOSBox-X mixer.
-
-
Default: 48000
-
Possible values: 44100, 48000, 32000, 22050, 16000, 11025, 8000, 49716
Sample rate in Hz of the MT-32 emulation.
-
Default: 2
-
Possible range: 0-3
MT-32 sample rate conversion quality
-
'0' is for the fastest conversion
-
'3' provides for the best conversion quality.
-
Default: true
-
Possible values: true, false
Toggles "Nice Amp Ramp" mode that improves amplitude ramp for sustaining instruments. Quick changes of volume or expression on a MIDI channel may result in amp jumps on real hardware. When "Nice Amp Ramp" mode is enabled, amp changes gradually instead. Otherwise, the emulation accuracy is preserved.
Unlike with the MT-32 emulation, General MIDI emulation is not implemented in DOSBox-X itself, but rather provided by a separate software synthesizer. DOSBox-X has support for a variety of such software synthesizers.
Note
|
There are also the Roland GS extensions to General MIDI, as used on the Roland Sound Canvas devices like the SC-55, SC-88 and SCC-I. Some games specifically list "Sound Canvas" as an option, but most of the time it just uses the same driver as if you select General MIDI. Games known to actually support the GS extensions can be found here: Sound Driver Enhancement Hacks - General MIDI (archive.org link, as a recent Vogons update has broken rendering of coloured text). |
FluidSynth is the preferred software synthesizer for Linux, but it is also available for Windows and macOS hosts.
DOSBox-X provides a range of configuration options, but for most Linux systems you can get it up and running simply by installing a SoundFont, from the distro package manager, such as "fluid-soundfont-gm".
If you installed DOSBox-X using one of the provided RPM packages, this SoundFont will be automatically installed by your package manager.
And then simply adding the following lines to your DOSBox-X config file:
[midi] mididevice=fluidsynth
In some cases, you may also need to specify a SoundFont file with fluid.soundfont=
, and the sound server with fluid.driver=
FluidSynth support is included in both Visual Studio and MinGW builds by default in DOSBox-X version 0.83.7 and later. You can add the following lines to your DOSBox-X config file.
[midi] mididevice=fluidsynth fluid.soundfont="C:\DOSBox-X\soundfonts\FluidR3_GM.sf2"
Adjust the path and filename to your SoundFont as necessary (e.g. "C:\DOSBox-X\GeneralUser_GS.sf2" instead of "C:\DOSBox-X\soundfonts\FluidR3_GM.sf2"). When no soundfont is specified, DOSBox-X will try to open C:\soundfonts\default.sf2 if it exists.
-
Default value:
-
Linux: pulseaudio
-
macOS: coreaudio
-
Windows: dsound
-
-
Possible values: pulseaudio, alsa, oss, coreaudio, dsound, portaudio, sndman, jack, file, default
This parameter is typically not needed if you’re running FluidSynth on Windows, where it will automatically use dsound. Likewise on macOS it will default to CoreAudio.
You can however optionally set it to "file", which will cause a fluidsynth.wav
file to be created in the current working directory with the MIDI output.
For Linux, the "pulseaudio" (default) and "jack" options are for different Sound Servers. Most modern Linux distributions by default install PulseAudio. Jack is lower latency, but higher CPU. It is possible to replace an installed PulseAudio Sound Server with Jack, but audio output of many applications will stop working, including web browsers. An alternative if you want to use Jack for FluidSynth is to set up PulseAudio as a client of Jack, but the setup of that is beyond the scope of this guide. It is expected that "PipeWire" will replace both PulseAudio and Jack in the near future.
For Windows and macOS you can optionally download and install FluidSynth. Note that not every update is released as binary, so unless you can compile it yourself you may be a few versions behind.
-
Default value:
-
Windows:
C:\soundfonts\default.sf2
-
Other:
/usr/share/soundfonts/default.sf2
or/usr/share/sounds/sf2/FluidR3_GM.sf2
-
This parameter can be used to specify a single SF2 or SF3 SoundFont file.
Many sound fonts can be found online, but you want one that covers the whole General MIDI range, also called a "GM SoundFont". FluidR3_GM.sf2 is such a free SoundFont.
Tip
|
For some games, you may want to use a SoundFont that implements both GM and the GS extensions, or a SoundFont that mimics a certain MIDI sound module. An example is the FluidR3 GM+GS SoundFont, which has both GM and the GS extensions in a single file. Alternatively, you can look for a SoundFont that mimics a certain MIDI sound module like the Roland SC-55 or SC-88, but all the Sound Canvas look-alike sound fonts available at present don’t seem to implement the GS extensions, and as such are only useful for GM. The Roland SC-55 or SC-88 (Sound Canvas) in particular, was used by many musicians back in the day to compose General MIDI music for DOS and Windows games. |
Note
|
While FluidSynth supports stacking or chaining of sound fonts, this is not supported in this implementation. So for instance, you cannot specify both FluidR3_GM and FluidR3_GS. Instead, you need a single SoundFont that implements all the sounds you need, which the above linked "FluidR3 GM+GS" does. |
-
Default value: 48000
-
Min - Max: 8000.0 - 96000.0
This effects the sample rate at which FluidSynth outputs audio.
Any modern system should support the default 48000 Hz.
You should set this to the same sample rate as the rate=
option in the [mixer]
section.
-
Default value: .6
-
Min - Max: 0.0 - 10.0
This value effects the output volume level of FluidSynth. If you experience that your background MIDI volume makes it impossible to hear voices in a game or other sound effects, you will want to lower this value. In such cases 0.2 seems to be a good value.
-
Default value: 256
-
Min - Max: 1 - 65535
This effects how many voices can be played in parallel.
-
Default value: default
-
Min - Max: 1 - 256
By default, fluidsynth will use a single CPU core. If you set this value higher, fluidsynth will create additional synthesis threads.
-
Default value: 16 (Linux and macOS)
-
Default value: 8 (Windows)
-
Min - Max: 2 - 64
The number of the audio buffers used by the driver. This number of buffers, multiplied by the buffer size (see setting fluid.periodsize), determines the maximum latency of the audio driver.
-
Default value: 64 (Linux and macOS)
-
Default value: 512 (Windows)
-
Min - Max: 64-8192
The size of the audio buffers (in frames).
-
Default value: yes
When set to "yes" the reverb effects module is activated. Otherwise, no reverb will be added to the output signal. Note that the amount of signal sent to the reverb module depends on the "reverb send" generator defined in the SoundFont.
-
Default value: yes
When set to "yes" the chorus effects module is activated. Otherwise, no chorus will be added to the output signal. Note that the amount of signal sent to the chorus module depends on the "chorus send" generator defined in the SoundFont.
-
Default value: .61
-
Min - Max: 0 - 1
Sets the room size (i.e. amount of wet) reverb.
-
Default value: .76
-
Min - Max: 0 - 100
Sets the stereo spread of the reverb signal.
-
Default value: 1.2
-
Min - Max: 0 - 10
Specifies the output amplitude of the chorus signal.
-
Default value: 8.0
-
Min - Max: 0 - 256
Specifies the modulation depth of the chorus.
TiMidity++ is a MIDI software synthesis, primarily for Linux. You should consider using FluidSynth instead. But if you want to try TiMidity++, install the timidity++ package with your package manager, and also install a SoundFont like FluidR3_GM.
You then need to point TiMidity++ to the SoundFont you want to use, for this edit /etc/timidity++/timidity.cfg
and make sure the soundfont
setting points to a valid SF2 or SF3 SoundFont file.
TiMidity++ support is not built-in like FluidSynth, instead you need to run it as a separate background service to which DOSBox-X can connect. After you have installed the packages, enable and start the timidity service. This only needs to be done once. On a modern Linux distribution this can be done as follows from the command line:
sudo systemctl enable timidity
sudo systemctl start timidity
Before trying MIDI in DOSBox-X, validate that TiMidity++ is working by playing a MIDI file.
timidity example.midi
Once you confirmed that MIDI works, set the below options in your DOSBox-X config file. This will cause DOSBox-X to try to connect to TiMidity++ over TCP/IP to localhost (127.0.0.1) on port 7777:
[midi]
mididevice=timidity
Using midiconfig=
a different host and/or port can be specified:
[midi]
mididevice=timidity
midiconfig=localhost:8000
Alternatively it is also possible to connect to TiMidity++ over ALSA as follows.
First locate the ALSA MIDI device that got assigned to TiMidity++ using aconnect -i
(in the Linux terminal) or mixer /listmidi alsa
(in the command line of DOSBox-X 0.83.12 or later). Then use the following config options:
[midi]
mididevice=alsa
midiconfig=128:0
ALSA is the low-level Linux sound system. You can use ALSA if you have a real MIDI sequencer, or a sound card with MIDI support. You can optionally also use ALSA with a separate software sequencer that creates a ALSA MIDI interface, such as TiMidity++ or Munt.
In DOSBox-X 0.83.12 or later you can use DOSBox-X’s built-in MIXER command to find which "device" is your MIDI output, e.g.
mixer /listmidi alsa
Alternatively, you can use the command aconnect -i
in the Linux terminal to find which "device" is your MIDI output.
Once you get the device ID (e.g. 128:0) you can then add it to the midiconfig=
line.
E.g.
[midi]
mididevice=alsa
midiconfig=128:0
OSS is an older low-level sound system for Unix and Linux. It should not be used if you have ALSA. You can use OSS if you have a real MIDI sequencer, or a sound card with MIDI support.
TBD.
This is the default MIDI emulation option on Windows when using mididevice=default
or mididevice=win32
(note: the "win32" option naming is a misnomer as it applies to both 32 and 64bit versions of Windows).
The advantage is, that it requires no configuration. The disadvantage is that it has high-latency and many of the sounds don’t sound correct. On a modern Windows system, there are no configuration options either.