Falcosoft wrote:

No. […]
Show full quote

Do any games actually use this special dual-synth mode, or more channels than vanilla Munt offers?

No.

That if a game supports general midi, but not mt-32, it will work?

Yes.

Why would someone want to send GM to an mt-32 emulator? I'm sure I am simply not understanding this correctly.

Good question (Where's Roland User when he's needed? ) 😀
There are people who desperately wanted this. Look at the SC-VA thread where you come from:
Heads Up: Roland Sound Canvas VA VSTi Plugin!
Heads Up: Roland Sound Canvas VA VSTi Plugin!
Anyway the special sound/mood of Munt/MT-32 makes it truly interesting to listen some GM midis on it. Strictly not for purists!
Try e.g. some Warcraft 2 midis. I really like how they sound.
Roland itself released an official pack in 1993 that made the MT-32 somewhat GM compatible ( but of course only 9 channels out of 16 could be used).

But obviously there are more legitimate scenarios where it can be useful/interesting.
Since it's a VSTi plugin its use is not restricted to retro gaming.

In general (and in layman's terms, preferably), why should I consider using the Munt VSTi over vanilla Munt?

Any advantages that a VSTi plugin offers... (used together with effects, in a DAW, etc.)
For retro gamers like you:
1. It's portable, so no installation is required (works just like SYXG-50) .
2. If you use a VSTHost anyway you do not have to change DOSBox's output to switch between e.g. SYXG-50, SC-VA and Munt.
3. Recording the output and changing some settings may be more straight forward than on standard Munt.
+ It works on Win9x 😀

Thank you very much for the detailed answers. I was wondering if you could perhaps go into a bit more detail on point number 2. I.e., You say that DOSBox's output does not have to be changed to switch between different VST synths -- are you referring to not having to change the mididevice parameter of DOSBox? In other words, LoopMIDI must be installed and once that one port is set as the mididevice in DOSBox (and in FSMP), the synths can be changed in FSMP either in the program or by sysex messages, correct? So the benefit of this is consolidating everything into one LoopMIDI port and changing the devices within FSMP instead of each device having its own LoopMIDI port.

Also, in playing around with Munt VSTi and the most recent DOSBox ECE build today, I noticed something. Munt VSTi seems to have some distortion (aliasing?) when I play King's Quest IV during the high trumpet notes of the intro. Munt VSTi also sounds louder in general than DOSBox ECE's built-in Munt emulation. In fact, I had to increase the windows speaker volume by 25 to match the volume level of DOSBox ECE built-in Munt with that of Munt VSTi. It is significantly louder.

Here's a comparison video I made of the vanilla Munt emulation in DOSBox ECE SVN r4019 vs Munt VSTi.

In that video, if you pay attention to the small speaker icon near the clock at the bottom-right of the desktop, you will see how the volume decreases once the Munt VSTi clip starts playing, and yet it sounds as loud if not louder than the first clip with the built-in Munt emulation.

Here are my Munt VSTi settings. They should be pretty much default:

Any ideas what might be causing this distortion?

Sounds like one is too loud and is clipping audibly.

kode54 wrote:

Sounds like one is too loud and is clipping audibly.

Right, it's the Munt VSTi one that has the clipping/distortion/volume issue. But, turning the output and reverb gain down doesn't seem to help, plus, it doesn't remember the setting (decreased gain settings) the next time munt vsti is launched.

Try the floating point processing option?

kode54 wrote:

Try the floating point processing option?

Hmm, didn't seem to make any noticeable difference. I still get the distortion.

From Falcosoft's website, I noticed this under the notes for the most recent version of Munt VSTi:

Added some explantion to DAC input mode selection about how different modes influence volume level.

I've been trying to find this documentation but so far have been unable to. DAC input mode options are as follows:

High Quality
Pure
Generation 1
Generation 2

High quality seems to be the default selection. However, I don't see any documentation on how these modes affect things.

Edit: Wonder if this is the same thing. Found this in DOSBox ECE r4019 config file:

# mt32.dac: MT-32 DAC input emulation mode # Nice = 0 - default # Pro […]
Show full quote

# mt32.dac: MT-32 DAC input emulation mode
# Nice = 0 - default
# Produces samples at double the volume, without tricks.
# Higher quality than the real devices
#
# Pure = 1
# Produces 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 any of the other modes.
# Perfect for developers while debugging 😀
#
# GENERATION1 = 2
# Re-orders the LA32 output bits as in early generation MT-32s (according to Wikipedia).
# Bit order at DAC (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).
# Bit order at DAC (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
# Possible values: 0, 1, 2, 3.

It looks like "Nice" and "High Quality" may be the same? If that's so then the built-in Munt of ECE should also be louder as well, it would seem.
(Edit: tried mt32.dac=1 with built-in Munt of ECE and it was indeed quieter than it already was)

Edit #2: After playing around some more, I have found that choosing "Pure" (DAC input mode) in Munt VSTi does get rid of the distortion, and lowers the volume slightly. But, if I then enable floating point, the volume goes back up and the distortion comes back, even with Pure still selected.

So, in short, it seems that between built-in Munt of ECE and Munt VSTi, Munt VSTi produces distortion when mt32.dac=0 (assuming that this is equivalent to the "High Quality" DAC input mode selection in VSTi) whereas the built-in Munt of ECE does not. Munt VSTi also seems inherently louder.

lukeman3000 wrote:

Edit #2: After playing around some more, I have found that choosing "Pure" (DAC input mode) in Munt VSTi does get rid of the distortion, and lowers the volume slightly. But, if I then enable floating point, the volume goes back up and the distortion comes back, even with Pure still selected.

So, in short, it seems that between built-in Munt of ECE and Munt VSTi, Munt VSTi produces distortion when mt32.dac=0 (assuming that this is equivalent to the "High Quality" DAC input mode selection in VSTi) whereas the built-in Munt of ECE does not. Munt VSTi also seems inherently louder.

Falcosoft and I have been discussing that on page 19: link

I think I will create a test MID for the heardware MT-32 to measure its signal-to-headroom ratio and compare with Munt32 and Munt VSTi.

James-F wrote:

lukeman3000 wrote:

Edit #2: After playing around some more, I have found that choosing "Pure" (DAC input mode) in Munt VSTi does get rid of the distortion, and lowers the volume slightly. But, if I then enable floating point, the volume goes back up and the distortion comes back, even with Pure still selected.

So, in short, it seems that between built-in Munt of ECE and Munt VSTi, Munt VSTi produces distortion when mt32.dac=0 (assuming that this is equivalent to the "High Quality" DAC input mode selection in VSTi) whereas the built-in Munt of ECE does not. Munt VSTi also seems inherently louder.

Falcosoft and I have been discussing that on page 19: link

I think I will create a test MID for the heardware MT-32 to measure its signal-to-headroom ratio and compare with Munt32 and Munt VSTi.

Hmm.. After reading through those posts I'm not sure I was able to ascertain if Falcosoft considers this issue resolved or not, or if he acknowledges it as an issue. Much of this is over my head, so I'm just trying to read between the lines.

As an aside, what do you recommend for the mt32.dac parameter for the "best" experience, and why?

For now I suggest the 'Pure' setting, but I would like to confirm the amount of headroom in the actual hardware MT-32.
SC-VA has plenty of headroom, so much so that users complain about it being too quiet, but actually it has exactly the same amount of headroom as the hardware SC-55.

Hardware MT-32 headroom is still a mystery but if anyone with a MT-32 willing to help and record the output of the MT-32 test tones I've attached.

I attached a zip with four files:
MT-32 Test Tone.MID
MT-32 Headroom.MID
GS Test Tone.MID
GS Headroom.MID

The MT-32 MID files send sysex messages to set the MT-32 unit before the test tone (mt-32 reset, reverb off, timbre change).
'Headroom' will send a lot of notes together with maximum velocity and patch volume with the intention to clip the internal op-amps of the MT-32 or GS unit, be careful not to clip the following input you record/mix into.
'Test Tone' is a stable tone that should provide a reference in contrast to the 'headroom' tone.

If someone with the hardware MT-32 (or similar like CM-32 etc..) willing to record the tests please do the following:
1. Play 'Headroom.mid' with FSMP and adjust your recording input so they don't clip, but the sound may already be distorted from the MT-32.;; be careful it's loud!
2. Record a few seconds of 'MT-32 Headroom.MID'.
3. Now continue recording a few seconds of 'MT-32 Test Tone.MID' WITHOUT changing anything on the MT-32.
4. Upload the single resulting WAV file here zipped.

This should give us a pretty clear picture of how much headroom the hardware MT-32 has.

@lukeman3000:
Please, sometimes just take a look at other posts in this thread. Just before your posts, I posted this:
Falcosoft Soundfont Midi Player + Munt VSTi + BassMidi VSTi
relevant part:

version 4.7 of FSMP and version 2.1 of MuntVSTi are released ... MuntVSTi 2.1 1. Added implementation of Pure DAC input mode in […]
Show full quote

version 4.7 of FSMP and version 2.1 of MuntVSTi are released
...
MuntVSTi 2.1
1. Added implementation of Pure DAC input mode in case of floating point rendering.
2. Added some explantion to DAC input mode selection about how different modes influence volume level.

So the problem is you still use version 2.0 of MuntVSTi (as seen on your screenshot).
Plese download the newest one (version 2.1). It's also in Midiplayer 4.7 package.

But, turning the output and reverb gain down doesn't seem to help, plus, it doesn't remember the setting (decreased gain settings) the next time munt vsti is launched.

The settings are saved by the host and not the plugin. So in your case you have to set SAVIHost to save the plugin's settings automatically.
SAVIHost-> File-> Autosave Plugin Bank. I have just checked and it works as it should (even worked in version 2.0).
In FSMP you have to save the setting manually by 'Save' and check 'Autoload last used settings File'.

Edit:
a video about using it:
https://youtu.be/QZeHHcxOJ8Y

@James-F and kode54:
Guys, you are experienced and knowleadgeable, please help to find out what is the situation with the 'proper volume level' compared to hw MT-32.
Facts:
1. I use unmodified version of Munt library (mt32emu) in MuntVsti and do not modify the rendered samples any way in the plugin (they are sent 'as is' to the host)
2. There are 2 functions in Munt's C API to get the rendered samples (note: the 2nd one is not the same as the newly introduced floating point rendering path) :
a. mt32emu_render_bit16s()
b. mt32emu_render_float()
For a VSTi plugin mt32emu_render_float is the natural choice since VST plugins always use floating point data(32 or 64 bit).

The 2 functions give different results regarding volume level. The floating point one results in twice the volume.
Which is the right one?
(The driver uses the 16 bit integer one)

I have posted this question also in the Munt developer thread:
Re: Munt Reloaded - Development

Falcosoft wrote:

please help to find out what is the situation with the 'proper volume level' compared to hw MT-32.

Let me start by saying I don't have any form of hardware MT-32, so the only testing option I have is to ask for recordings of test midis.
So it's a black box that we're trying to find its transfer function with stimulus and response.. the usual.

Download MT-32 service manuals from here: http://www.synfo.nl/servicemanuals/Roland/MT- … E_NOTES-3rd.pdf
Looking at the block-diagrams and schematics I can see that it is typical with Roland synths to have amplifiers before the volume control and after it, both may distort.
The PRE volume control op-amps distort with high velocity midis, and the POST volume control op-amps distort with high volume control settings, this is true for both the MT-32 and SC-55.
High velocity midis and high volume control setting will result in multiple distortion stages, like in a guitar amplifier (Mesa Boogie, Marshall, Bogner etc,).. 😀

As I understand the GAIN controls in MUNT are like the PRE volume gains which are fixed by resistors in the hardware.
Hardware: DAC -> PRE Amplifier (fixed gain) -> Volume Control -> POST Amplifier -> Output.
Munt: DAC -> PRE Amplifier (adjustable Gain) -> Volume Control (FSMP or Windows) -> Output.

In the digital domain the POST volume control limit is the digital headroom of 0.0dbFS since the VST plugin itself uses floating point as you say.
The PRE volume control headroom is dictated by "emulated" or pre-programmed by the "mt32emu_render_bit16/32()" line you talk about, which is our goal to find our.
In contrast to the accurate emulation of MUNT, SC-VA has no "emulated" headroom, it is limitless unlike the HW SC-55 or MUNT.

So we need a recording of the "MT-32 headroom.MID" and "MT-32 Test Tones.MID" I've provided two posts ago to figure what the actual PRE volume control headroom of the HW MT-32.
The recording of the hw MT-32 has to be done with the volume pot at low setting to not clip the POST vol op-amps which we don't care about (yet) in this test.
Then we can match the proper MUNT emulation DAC mode and know the proper Gain setting for the emulated DAC.

This is as far as I can be helpful without actually owning the hardware.

I've PM'ed Phil, hope he chimes in.

Sorry just not a good time at the moment. All my MIDI gear is packed away, I'm using that Windows tablet with USB MIDI Adapter because it's less clutter and simpler.

I will be doing DOS projects / videos in maybe a month or two again, but at the moment I simply do not have the time I'm afraid.

Alright thanks Phil, I agree that plugins are much more convenient, that's why we are trying to make them as accurate as possible. 😀
Hmm.. who else on Vogons has a bunch of MT-32 hardware that can do a proper recording?

Falcosoft wrote:

@lukeman3000: Please, sometimes just take a look at other posts in this thread. Just before your posts, I posted this: Falcosoft […]
Show full quote

@lukeman3000:
Please, sometimes just take a look at other posts in this thread. Just before your posts, I posted this:
Falcosoft Soundfont Midi Player + Munt VSTi + BassMidi VSTi
relevant part:

version 4.7 of FSMP and version 2.1 of MuntVSTi are released ... MuntVSTi 2.1 1. Added implementation of Pure DAC input mode in […]
Show full quote

version 4.7 of FSMP and version 2.1 of MuntVSTi are released
...
MuntVSTi 2.1
1. Added implementation of Pure DAC input mode in case of floating point rendering.
2. Added some explantion to DAC input mode selection about how different modes influence volume level.

So the problem is you still use version 2.0 of MuntVSTi (as seen on your screenshot).
Plese download the newest one (version 2.1). It's also in Midiplayer 4.7 package.

Hi Falcosoft -

Indeed I both saw AND read that post (which in fact was how I was able to quote it in my own). However, this time, it was not an issue of looking, but rather failing to realize that I was not using the most recent version of Munt VSTi, which would explain why I was unable to find the documentation. Thank you for bringing this to my attention.

@lukeman3000:

I realize that I seem to be one of the less intelligent people here on vogons...

I don't think I have suggested such a thing. I have just noticed that you are sometimes looking for information that is already available in earlier posts. That's all.

aggravate anyone... I don't piss anyone off here...

I don't feel the 'anger' in my post that may make you feel you have pissed me off. If you do so, that was not my intention. (Usually I do not make tutorial videos about an issue when I'm pissed off...)

Ps: Have you solved your 'Gain settings not saved' issue?

Falcosoft wrote:

@lukeman3000: […]
Show full quote

@lukeman3000:

I realize that I seem to be one of the less intelligent people here on vogons...

I don't think I have suggested such a thing. I have just noticed that you are sometimes looking for information that is already available in earlier posts. That's all.

aggravate anyone... I don't piss anyone off here...

I don't feel the 'anger' in my post that may make you feel you have pissed me off. If you do so, that was not my intention. (Usually I do not make tutorial videos about an issue when I'm pissed off...)

Ps: Have you solved your 'Gain settings not saved' issue?

Sorry, I was frustrated for other reasons unrelated; I already have removed that part of my previous post.

As far as the gain settings - I'll have to wait until I get off work to do some more testing.

James-F wrote:

For now I suggest the 'Pure' setting, but I would like to confirm the amount of headroom in the actual hardware MT-32.
SC-VA has plenty of headroom, so much so that users complain about it being too quiet, but actually it has exactly the same amount of headroom as the hardware SC-55.

Hardware MT-32 headroom is still a mystery but if anyone with a MT-32 willing to help and record the output of the MT-32 test tones I've attached.

Here are recordings on a CM-64 at 90 and 100 percent line-in levels.

I can also provide MT-32 (Old) recordings if it matters.

Thank you very much HunterZ.

Line-In levels in windows? What about the volume pot on the front panel of the CM-64?
It seems that at 100 and 90 the input of the capturing sound card are clipping (or are they normalized?).
If you can, please redo the recordings without normalizing or clipping the input of the sound card, and in the lower quarter of the CM-64 volume control (10 o'clock max).
Nothing should reach 0.0dbFS in your recording software.

Yes, MT-32 (old) recordings will be excellent to have as reference.

No normalizing, so it's certainly just clipping at the CM-64. I usually don't mess with the volume knob on the LA synths because I think SysEx can override them anyway? I'll try adjusting the CM-64 knob and make another recording, and then maybe make a similar go with the MT-32.

Not normalized?
The WAV files reach 0.0dbFS, that means either the recording input is clipping (please lower Line-In levels) or normalized to 0.0dbFS.
Yes, please redo the recording with the Line-In levels low enough so that the recorded sound of the CM-64 does not peak the reds in your recording software.
And use the lower quarter of the volume pot on the front panel, even if SysEx re-writes it.

The volume control on the MT-32 unit is a VCA (voltage controlled amplifier) controlled by PWM from the CPU so its not a variable resistor, that's why a SysEx message can overwrite it.