Topic: XMPlay MIDI plugin

[piovrauz]

Jan, U rock!

Thanks for the automatic loading on matching filenames I requested, and for the fix for bad bank management, pointed by another user: I was having the same thing, you corrected this in no time!

I'm enjoing my MIDI collection again! Now I can use sf2 in the easy way, without bloated players, and get a great sound! I'm listening non stop all my MIDI, that I kept for better days.. And the day is now!

But since I noticed your speed in fixing bug/errors, I have one report too: in a MIDI I got with my old AWE64G a sample in the sf2 is played at wrong speed and cutted after some time. Glitch is at time 01:38, I suppose it's track 07, the "extra melody".

I'll try to upload the MIDI and the sf2.

Edit: FTP doesn't like me... Sended email to admin (at) un4seen (dot) com.

[Ian @ un4seen]

(Edit) Just had a look at the MIDI file in question in my sequencer, it appears that the lyrics are tagged as text events, and not lyric events. Strange thing is, the words display properly in Tune 1000 (as well as another karaoke player that I used to use... don't remember the name offhand). Could this be why the MIDI plugin doesn't display the lyrics?

Yep, that's why I specifically mentioned "proper lyrics event"

I've added support for these general text lyrics things now too.

(Edit #3 - re: The "sndfont.c" source file mentioned/linked above) Here is the "sbkconv.c" source code file referenced by that other file -> http://www.csee.umbc.edu/help/sound/TiMidity++-2.13.2/timidity/sbkconv.c

Looks a bit messy. There was never an official SBK spec released, was there? Not sure I'll bother with it, who/why use SBK anyway?

...I have one report too: in a MIDI I got with my old AWE64G a sample in the sf2 is played at wrong speed and cutted after some time. Glitch is at time 01:38, I suppose it's track 07, the "extra melody".

This seems to be a problem with the MIDI file not playing the note long enough, or the soundfont not sustaining the sample. Either way, it sounds the same played on SB (Audigy2) hardware

Maybe the SF2 used to be an SBK, and wasn't converted properly?

[Rich Nagel]

I've added support for these general text lyrics things now too.

Works perfectly, many thanks! :thumbsup: -:)

Looks a bit messy. There was never an official SBK spec released, was there? Not sure I'll bother with it, who/why use SBK anyway?

Yeah, I don't think that there was any officially released specs for the SBK format. I searched all over my PCs (as well as the Internet) and came up with nothing.

As you said, there probably wouldn't be much of a demand for SBK support anyway. Hehe, the reason that I mentioned it was that I have a lot of (other composers) creations that use custom SBK files for specific MIDI songs (most of them are on almost all of the driver CDs that ship with the AWE32/AWE64 sound cards, one that comes to mind is "Mystic Dragon"... I think that was the name of the tune). Although quite dated (they all use SBKs), they sound rather nice.

Anyhow, one could always use the SF1TO2.EXE DOS command-line converter utility to convert the SBK files to SF2 format, after experimenting with it, it seems to work fine -:)


P.S. Gotta say it again, THANKS FOR THE ROCKIN' SF2 MIDI SUPPORT FOR XMPLAY! :thumbsup: -:)

I remember wishing for something like this *eons* ago, back when I wanted others to hear my MIDI music compositions... *exactly* the way that *I* heard them on *MY* PC <G>, but they didn't have any SF2 compatable hardware. Hehe, this was before the popularity of the MP3 format, so (on a miserably poor dialup connection) extremely low quality WAVs were the only solution <G>.

Years later Timidity came along, but I was never happy with the resulting sound of it (and some SF2s really sounded strange when used as the base sound bank).

[Zarggg]

The FluidR3 GS soundfont causes XMPlay to crash for me.

Is that the one I posted above, or a different version? If it's the one above, please let me know if there's a particular MIDI file that triggers the crash, as I was unable to get it to crash. A drwtsn32.log entry would be handy too.

It's the one you posted. I downloaded the new beta and it no longer crashes. However, I don't hear any sound with that specific soundbank, either.

That's strange. Have you un-sfArk'd the file? What's the size of your "FluidR3 GM.SF2" file?

"FluidR3 GM.SF2" is 148,358,590 bytes.
"FluidR3 GS.SF2" (the one I was talking about) is 3,200,570 bytes.

[Rich Nagel]

@Ian, (trivial) feature request: This is rather trivial (but I would use it quite a bit <G>), adding a (possibly configurable) hotkey to bring up the MIDI plugin's configuration window? 'Twould be handy for quickly changing the base SF2s (as well as any other features that may be added within that window, i.e. multiple stacked SF2s, etc...).

(Edit) Another small feature request (Sorry, Ian! I'm just quite excited by the capabilities of this plugin <G>!): Some SF2s contain pointers to onboard AWE32/AWE64 ROM samples (which, of course, are not actual samples contained in the SF2 file, but rather pointers to samples contained in the AWE32's/AWE64's onboard ROM chip).

Would it be possible when autoloading an SF2 with a matching MIDI filename (as well as stacked soundfont support, if you add that feature) for the plugin to use the samples contained in the base SF2 - if the custom SF2 contains a pointer to a ROM sample/preset? Maybe something similar to how the plugin already uses presets from the base SF2 if one is not already contained in the custom SF2 (but in the afore-mentioned example, the preset is actually there - a pointer to an AWE32/AWE64 onboard ROM sample, but the actual digital samples aren't contained in the SF2).

[Ian @ un4seen]

The FluidR3 GS soundfont causes XMPlay to crash for me.

Is that the one I posted above, or a different version? If it's the one above, please let me know if there's a particular MIDI file that triggers the crash, as I was unable to get it to crash. A drwtsn32.log entry would be handy too.

It's the one you posted. I downloaded the new beta and it no longer crashes. However, I don't hear any sound with that specific soundbank, either.

That's strange. Have you un-sfArk'd the file? What's the size of your "FluidR3 GM.SF2" file?

"FluidR3 GM.SF2" is 148,358,590 bytes.
"FluidR3 GS.SF2" (the one I was talking about) is 3,200,570 bytes.

Oh right. That soundfont only contains extra effects banks/patches, ie. there's no bank 0. I guess it's meant to be stacked on top of a GM soundfont.

@Ian, (trivial) feature request: This is rather trivial (but I would use it quite a bit <G>), adding a (possibly configurable) hotkey to bring up the MIDI plugin's configuration window? 'Twould be handy for quickly changing the base SF2s (as well as any other features that may be added within that window, i.e. multiple stacked SF2s, etc...).

I'm afraid that's not currently possible, as plugins can't register their own shortcuts.

Would it be possible when autoloading an SF2 with a matching MIDI filename (as well as stacked soundfont support, if you add that feature) for the plugin to use the samples contained in the base SF2 - if the custom SF2 contains a pointer to a ROM sample/preset? Maybe something similar to how the plugin already uses presets from the base SF2 if one is not already contained in the custom SF2 (but in the afore-mentioned example, the preset is actually there - a pointer to an AWE32/AWE64 onboard ROM sample, but the actual digital samples aren't contained in the SF2).

Please upload an example of this, and I'll have a look

[Rich Nagel]

I'm afraid that's not currently possible, as plugins can't register their own shortcuts.

That's what I figured, thanks for the info -:)

Please upload an example of this, and I'll have a look

I uploaded rom_samples_and_user_samples_test.zip (contains "ROM Samples and User Samples Test.sf2").

The SF2 contains all AWE32/AWE64 ROM samples for GM presets except for:

(Melodic Preset)
22 = Harmonica

(Percussive Preset)
16 = Power (Note: *NOT* a complete power drum set, only contains samples/instruments for kick, snare, and toms - *NO* pointers to AWE32/AWE64 ROM samples for the rest of the kit).

[Zarggg]

(snip cascade :p)

Oh right. That soundfont only contains extra effects banks/patches, ie. there's no bank 0. I guess it's meant to be stacked on top of a GM soundfont.

I'll accept that answer. I've settled on using the Chorium SF anyway.

[Rich Nagel]

@Ian, I think I may have found a small buglet with the way that the plugin handles the "Modulation LFO" envelope/parameter within any given soundfont (there's a strange badly distorted sound). I uploaded "xmplay_midi_plugin_modulation_lfo_bug.zip" to the FTP site (contains "XMPlay MIDI Plugin Modulation LFO Bug.sf2").

The SF2 has two presets in it, Tremelo Strings (44) and String Ensemble (48), and they both use the exact same samples and envelopes. The only difference between the two is the usage of the modulation LFO as configured/detailed below:

Tremelo Strings = 44
--------------------
Delay - 1 X (default)
Frequency - 10 X
To Pitch - 2 cents
To Filter Cutoff - 0 cents (default)
To Volume - 5 dB

String Ensemble = 48
--------------------
Delay - 1 X (default)
Frequency - 1 X (default)
To Pitch - 0 cents (default)
To Filter Cutoff - 0 cents (default)
To Volume - 0 dB (default)

The String Ensemble sounds correct, but the Tremelo Strings have the garbled distorted sound.

[Ian @ un4seen]

I uploaded rom_samples_and_user_samples_test.zip (contains "ROM Samples and User Samples Test.sf2").

Ok, I've added ROM detection, but I've not really tested it, so let me know if it's no good

@Ian, I think I may have found a small buglet with the way that the plugin handles the "Modulation LFO" envelope/parameter within any given soundfont (there's a strange badly distorted sound). I uploaded "xmplay_midi_plugin_modulation_lfo_bug.zip" to the FTP site (contains "XMPlay MIDI Plugin Modulation LFO Bug.sf2").

The SF2 has two presets in it, Tremelo Strings (44) and String Ensemble (48), and they both use the exact same samples and envelopes. The only difference between the two is the usage of the modulation LFO as configured/detailed below:

Tremelo Strings = 44
--------------------
Delay - 1 X (default)
Frequency - 10 X
To Pitch - 2 cents
To Filter Cutoff - 0 cents (default)
To Volume - 5 dB

String Ensemble = 48
--------------------
Delay - 1 X (default)
Frequency - 1 X (default)
To Pitch - 0 cents (default)
To Filter Cutoff - 0 cents (default)
To Volume - 0 dB (default)

The String Ensemble sounds correct, but the Tremelo Strings have the garbled distorted sound.

I guess you're comparing with SB hardware? The difference seems to be due to hardware limitations, as the Tremelo Strings should be modulating at 50hz (10x5), but it's only going at 12hz (on an Audigy2). The SF2 specs say the limit is 100hz.

I've now limited the LFO parameters to match the apparent SB hardware limits.

[Rich Nagel]

Ok, I've added ROM detection, but I've not really tested it, so let me know if it's no good

Works perfectly! Many thanks! -:)

I guess you're comparing with SB hardware? The difference seems to be due to hardware limitations, as the Tremelo Strings should be modulating at 50hz (10x5), but it's only going at 12hz (on an Audigy2). The SF2 specs say the limit is 100hz.

Actually, I tried it on an SB16 as well as a mobo with a Crystal Media sound chip, same results.

I've now limited the LFO parameters to match the apparent SB hardware limits.

Again, works perfectly, and again, many thanks! -:) :thumbsup:


Can't say it enough times <G>, thanks again for all of your hard work on this plugin! -:) -:)

[Maserati]

midi's never sounded so real   Thanks!

[Ian @ un4seen]

(snip cascade :p)

Oh right. That soundfont only contains extra effects banks/patches, ie. there's no bank 0. I guess it's meant to be stacked on top of a GM soundfont.

I'll accept that answer. I've settled on using the Chorium SF anyway.

To help identify situations like this, I've now added missing patches to the Samples/Instruments info.

[Rich Nagel]

To help identify situations like this, I've now added missing patches to the Samples/Instruments info.

Excellent feature, Ian, that's really handy :thumbsup: -:)

Also, I really need to try out that "Chromium" soundfont (I'm on a lowly dialup connection), as well as comparing it to (and maybe uploading) my own largish (excellent <IMHO> <G>) custom GM compatable soundfont -:)

[Rich Nagel]

Small feature request: (Actually, I think that this would probably be a request for XMPlay itself, dunno if it could be added to just the MIDI plugin)

Is there any way to add more reverb beyond the 100 percent that the slider witin the main XMPlay interface allows? Just an idea, possibly an "X" sort of checkbox, something like "Reverb Times (X) 2" (which would make the slider have twice the range: 0% - 200%).

For MIDIs, this would be similar to the *extremely* higher reverb effects that the AWE/EMU8000 hardware supports (compared to the standard reverb (#91) controller), when used with the NRPN and Data (MSB and LSB) MIDI controllers that EMU hardware supports.

As an example, adding the following sequence of MIDI controllers to a MIDI that is played back on AWE/EMU8000 hardware results in an insane amount of reverb <G> (*much* higher than the standard reverb MIDI controller #91):

99 NRPN MSB = 127
98 NRPN LSB = 26 (designates EMU8000 NRPN reverb - NRPN controller #26)
6 Data MSB = 127
38 Data LSB = 127

Anyhow, some sort of "times 2" checkbox for the reverb slider in XMPlay would simulate this higher reverb effect supported by the EMU hardware's NRPN 26 reverb controller.


Also, I was wondering how the stacked soundfont support was coming along? Note that I'm not trying to rush you Ian <G>, I'm simply extremely excited about this possible feature <BG> -:) Maybe a suggestion, a dialog box similar to how the AWE32 Control Panel looks (i.e. assigning and loading SF2s into various "user banks" - banks 1 through 255).


Lastly, I finally managed to download that Chorium soundfont on my lowly dialup connection <LOL>, and it's quite good -:) Although, after playing a few MIDIs, I must say that I *really* need to upload my custom General MIDI compatable soundfont when I get a chance <G>. A 35 meg or so ZIP file will be tedious on my dialup connection <LOL>, but I think that folks would like it -:)

[Rich Nagel]

The difference seems to be due to hardware limitations, as the Tremelo Strings should be modulating at 50hz (10x5), but it's only going at 12hz (on an Audigy2).

BTW, just for clarification, (in the SF2 that I uploaded) shouldn't that be 10hz (varying the pitch by 2 cents and the volume by 5 dB, at a frequency of 10 hz)?

[Ian @ un4seen]

Small feature request: (Actually, I think that this would probably be a request for XMPlay itself, dunno if it could be added to just the MIDI plugin)

Is there any way to add more reverb beyond the 100 percent that the slider witin the main XMPlay interface allows? Just an idea, possibly an "X" sort of checkbox, something like "Reverb Times (X) 2" (which would make the slider have twice the range: 0% - 200%).

Dunno, I'll have to think about it. Personally, I like the reverb level at a more subtle 40%

Also, I was wondering how the stacked soundfont support was coming along? Note that I'm not trying to rush you Ian <G>, I'm simply extremely excited about this possible feature <BG> -:) Maybe a suggestion, a dialog box similar to how the AWE32 Control Panel looks (i.e. assigning and loading SF2s into various "user banks" - banks 1 through 255).

I've not really started on it yet. The main use of stacking I had in mind was being able to use single patch soundfonts to replace patches that you don't like in the base soundfont. I was expecting/hoping that single patch soundfonts would be pre-assigned to the appropriate patch number, so that they'd be ready to go. But after downloading one, that's apparently not so - it's just assigned to patch 0. That complicates things somewhat. It means the config will have to allow the user to choose a patch number (not just bank) to assign a soundfont to.

The difference seems to be due to hardware limitations, as the Tremelo Strings should be modulating at 50hz (10x5), but it's only going at 12hz (on an Audigy2).

BTW, just for clarification, (in the SF2 that I uploaded) shouldn't that be 10hz (varying the pitch by 2 cents and the volume by 5 dB, at a frequency of 10 hz)?

Nope, the info you posted was from the "global zone", so that "10 X" means 10 times the instrument's frequency, which is 5hz... so 50hz it is

[Rich Nagel]

Dunno, I'll have to think about it. Personally, I like the reverb level at a more subtle 40%

Hehe, I always like to jack the reverb up extremely high, especially for classical/orchestral stuff <G> -:)

I've not really started on it yet. The main use of stacking I had in mind was being able to use single patch soundfonts to replace patches that you don't like in the base soundfont. I was expecting/hoping that single patch soundfonts would be pre-assigned to the appropriate patch number, so that they'd be ready to go. But after downloading one, that's apparently not so - it's just assigned to patch 0. That complicates things somewhat. It means the config will have to allow the user to choose a patch number (not just bank) to assign a soundfont to.

Yeah, many single patch SF2s aren't mapped with the correct preset/patch numbers. Some are, but some aren't... depends on who created them.

I was thinking more of where one could load in an SF2 in a specific user bank, and then if a given MIDI file had MIDI controller "bank select" events embeded in it (MIDI controller #0), it would use those specific banks for the specific tracks/channels within the MIDI.

If the stacking feature could allow one to assign the patch/preset number as well, that would be a cool feature in itself -:)

Nope, the info you posted was from the "global zone", so that "10 X" means 10 times the instrument's frequency, which is 5hz... so 50hz it is

Yes, you sir, are correct <G> -:) I had to dig up my old AWE composing notes to confirm... been a while since I created any custom soundfonts <G>.

[Ian @ un4seen]

An update with soundfont stacking is up now. It also now applies soundfont config changes mid-song (don't have to reload the file).

If there are no problems, I reckon the plugin is pretty much done now

[Rich Nagel]

An update with soundfont stacking is up now. It also now applies soundfont config changes mid-song (don't have to reload the file). If there are no problems, I reckon the plugin is pretty much done now

Had a look at the new features/options, it looks SUPER! So many new things to try, I'll experiment with it tonight and post a reply tommorow -:)

THANKS, IAN! :thumbsup: -:)

[Rich Nagel]

One word: AWESOME! -:) Everything works perfectly :thumbsup:, except for one little problem that I had. I uploaded power.zip to the FTP, it contains an example MIDI and the soundfont POWER.SF2.

The MIDI file is set up like this:

Channel 10 (GM percussion channel) - Bank 1 - Patch 16 (GM Power Drums patch)

and the SF2 has a few replacement samples to replace the kick, snare, and toms for patch #16 in the normally loaded base SF2 file. In other words, POWER.SF2 is not a complete set of drums, but rather only the kick, snare, and toms.

I couldn't seem to load in the POWER.SF2 file into user bank 1 (in order for it to be played by the example MIDI file). When I load it in the lower half of the config window (bank 128 in the upper half of the window, patch 16 in the lower half), you don't heard the rest of the kit when playing the MIDI file (you only hear the actual samples contained in the POWER.SF2 file).

I guess what I'm asking is, if a percussion SF2 is loaded that doesn't contain certain samples for that patch # (for the complete kit), can the rest of the kit be utilized from the matching patch/kit number in the base SF2?

Lastly, one last itty-bitty feature request <G>: A "Clear All" sort of button for the config window... would make things much easier when I was auditioning a bunch of MIDIs that used a bunch of different custom soundfonts, a different one for each MIDI channel, and each MIDI used totally different soundfonts <BG>. A "Clear All SoundFonts" button would be great -:)


Anyhow, superb job, Ian. Everything else works perfectly, with the new stacked soundfont support this thing is basically a "Virtual AWE32/AWE64/SBLive!/EMU-8000" <G> -:)


P.S. About the MIDI: something that I dreamed up when I was a young kid, and then many years later sequenced on the PC. Always reminded me of something that Fat Albert and the Gang would sing/preform in the old TV cartoon <G>, hence the filename of the MIDI <LOL>!

"HEY HEY HEY!... WHAT'S HAPPENIN' TODAY?"

<LOL>!

[Ian @ un4seen]

I couldn't seem to load in the POWER.SF2 file into user bank 1 (in order for it to be played by the example MIDI file).

Yep, bank changes weren't supported on drum channels (didn't know they were allowed ). I've added support for that now.

When I load it in the lower half of the config window (bank 128 in the upper half of the window, patch 16 in the lower half), you don't heard the rest of the kit when playing the MIDI file (you only hear the actual samples contained in the POWER.SF2 file).

This is trickier to "fix", but note that SB hardware doesn't fall back to other banks for individual keys either

Lastly, one last itty-bitty feature request <G>: A "Clear All" sort of button for the config window...

Done

[heftig]

The memory management is amazing. When I play a song (using a 1.6GB soundfont) XMPlay only needs 65MB of RAM, while Timidity eats up 450MB.

Unlike Timidity however, I cannot use XMPlay as a MIDI device when playing games and such.

[Rich Nagel]

Yep, bank changes weren't supported on drum channels (didn't know they were allowed ). I've added support for that now.

Works perfectly, thanks! -:)

This is trickier to "fix", but note that SB hardware doesn't fall back to other banks for individual keys either

Actually, on my AWE(s), it does. As an example, I can load in E-mu's 8MBGM.SF2 as the main SF2 on my AWE(s), and then load in the POWER.SF2 into user bank #1. When playing back MIDIs (that are mapped like the FALBERT2.MID example MIDI was: channel 10, patch #16, bank 1), I can hear the kick/snare/toms from the POWER.SF2, and the rest of the kit is retrieved/played from the 8MBGM.SF2 (contained in the "Power Drum" kit patch #16 in the 8MBGM.SF2).

Anyhow, if you can easily fix/add that feature it'd be great, but if not, I'm not complaining... this thing is great and AWEsome as it is -:)


(re: "Clear All" button)

Done

Superb! Thanks, Ian! :thumbsup: -:)

[Ian @ un4seen]

When I play a song (using a 1.6GB soundfont)...

 

Actually, on my AWE(s), it does. As an example, I can load in E-mu's 8MBGM.SF2 as the main SF2 on my AWE(s), and then load in the POWER.SF2 into user bank #1. When playing back MIDIs (that are mapped like the FALBERT2.MID example MIDI was: channel 10, patch #16, bank 1), I can hear the kick/snare/toms from the POWER.SF2, and the rest of the kit is retrieved/played from the 8MBGM.SF2 (contained in the "Power Drum" kit patch #16 in the 8MBGM.SF2).

That's strange. It doesn't do that here (on an Audigy2). Is there some special way of loading the soundfont to make that happen? I used the Soundfont Bank Manager v1.00.21.