gacki

Worth watching IMHO: https://www.youtube.com/watch?v=N8NyEjB_XeA

I did a few tests in 10.7.9. When opening a MIDI file the newly created track is correctly shown as a Software Instrument track right away (or more precisely: the track is not selected and clicking on it immediately shows it's a Software Instrument track). I also did a few tests with the "External Instrument" plugin on a Software Instrument track and it seems that this one DOES NOT SEND any SysEx. Bottom line: For sending SysEx we always need a dedicated "External MIDI" track, not a Software Instrument track with the "External Instrument" plugin.

Tests were done on Logic 10.5.1 but I can redo them on newer versions if the solution doesn't work. It's a really stupid bug with a totally simple solution. When importing a MIDI file Logic seems at first glance to create external MIDI tracks, allowing you to set the port and everything. But that's an illusion: In reality Logic has created a Software Instrument track, complete with Plugin chain and so on. But that only becomes obvious when creating a new track and going back to the imported track: So the track from the import is out of the box useless for sending data to an external instrument; the settings I changed are simply not there anymore. Solution: Use a newly created External MIDI track with the proper settings and everything should work.

Give me a bit of time; I'm certain I can crack this. I can partially reproduce the problem. Edir: I think I found the problem (and possibly the solution). Running further tests. Stay tuned.

Pretty close; it's a bit more elaborated. The two additional SysEx messages are one very short message that I've so far haven't been able to identify (it might be "Voice data receive block parameter change" - no idea what that actually does), and a second larger message with "Additional Voice Memory" aka AMEM. It's not the Performance stuff like the multitimbral setups; I believe AMEM is (in a nutshell) the Function parameters of the original DX7 but here on a per-voice basis instead of globally (the DX7II family did this as well if I recall correctly). I've run some more tests and Dexed doesn't accept anything other than a vanilla DX7 dump. A sensible way to handle this would be to scan through the whole .syx file to check if there actually IS some voice data instead of just jumping the gun and interpreting the first dump as "voice data" (in fairness: Dexed warns about this. But still...). It seems to be a known issue: https://github.com/asb2m10/dexed/issues/165 FM8 can actually import the combined dumps but seems to ignore AMEM. Same for the Arturia DX 7 V. So stripping down the files to the raw original DX7 SysEx seems to be the best course of action. Just reading through some of the TX802 SysEx documentation gave me vivid flashbacks...

I'm having a PowerMac 4400 for that exact reason. Back in the day I retrofitted it with an USB card which makes file transfers a bit easier - just use a thumb drive which can be read in both Mac OS 9 and current systems. While SoundDiver was starting up nicely and the library with the factory sounds was there it was always complaining about the missing MIDI interface (that's why I didn't sort the Patches). Exporting to .syx was no problem. But on my current Mac I couldn't import the .syx file into Dexed. Turns out this wasn't a single dump but a conglomerate of three different dumps (and different data types). But thanks to a hex editor I was able to extract the actual data which then could be imported. SoundDiver had been coming to the rescue a couple of times in the last few years. One time it was for restoring the factory patches to a Lexicon LXP-1 which had lost its memory due to a low battery. There was no .syx data for those patches to be found anywhere on the internet - but they were in the SoundDiver library. Another case was a Roland D-50 which had a battery change - but the owner sold it for little money because he wasn't able to restore the patches (since Roland at this time used a peculiar handshake protocol).

Here's the stuff. Sadly the files are not ordered correctly but I believe they contain all the patches. If not let me know. (As usual: SoundDiver to the rescue. SoundDiver came with a folder containing the factory patches for a number of devices including the TX802. The whole process of getting the stuff out of it and onto my current computer was somewhat messy though.) The files load in Dexed and FM8. TX802VoiceA2.syx TX802VoiceA1.syx

I'll see if I can find the TX802 Sysex elsewhere.

Those are from the Dexed cart; I have naturally no idea if they are all correct or work. TX802BankA1.SYX TX802BankA2.SYX TX802BankB1.syx TX802BankB2.syx

I would have to look through the Kronos manual for this; but here are a few quick thoughts: 1. Are you sure that those are CC's and not Program Changes? 2. At least control surfaces often require their own protocol to drive the LED's; those things are basically independent from the actual controllers. 3. Do you actually have a "backchannel" from Mainstage to the Kronos? In many cases (but certainly not all!) Mainstage is used mainly as the recipient for data coming from a controller but not automatically also as a sender back to the controller.

Jon Stewart made fun (or did shine a light on) about exactly this: "prompt engineers", as a Microsoft managers call them: https://www.youtube.com/watch?v=20TAkcy3aBY&t=574s But back to a more serious note: To me, creating music has to a certain degree always been about individuality and originality. I do most of my work in education nowadays; and I am always fascinated that people "pick" their favourite music at least partly "off the beaten path". I know they don't consciously pick it - there's simply something that resonates with them. So both the act of creation and the act of listening are both highly individualized processes: What speaks to me doesn't automatically speak to you and vice versa. And that's the beauty of it; and I am convinced that that's also a limit for AI.

I've tried numerous times to get something useful out of StaffPad but so far I've failed nearly every time. So I wouldn't hold my breath here. (IMO handwriting recognition still sucks for normal texts let alone music notation.)

I've listened to a couple of the "trending" songs featured on the starting page and to me they sounded rather bland and generic. Which isn't exactly surprising IMO because "originality" is comparatively hard/impossible to train. (We've been over this argument elsewhere before.) The musical segment of stock or library music will probably be most affected by this. Outside this we can only guess.

Yes, and written sheet music is in horizontal fashion. So at one point one has to make that mental switch anyway. The problem with the whole Synthesia piano thing is that it is basically hiding the "big picture". If you look at a piece of sheet music you can easily make out way more things at once. I blame that tendency on Guitar Hero and similar things.

It's a question of how we want time to be represented - horizontally like on a tape machine or vertically. It's always becomes slightly confusing when those two are mixed; and hence nearly all DAW's have switched to a horizontal scrolling. Youtubes piano tutorials of course use the vertical approach again which definitely drives me mad.