Qtractor not saving the MIDI info when exporting a MIDI

[Hoff123]

Yeah, anyway thank you for all your help. I'll probably post here again if I (ever) solve it. Bye .

[rncbc]

Yeah, anyway thank you for all your help. I'll probably post here again if I (ever) solve it. Bye .

please, don't take me wrong. i know something stinks i'm not dissing whether the wrong doing is on either side, whatsoever. i'm sure one of us is missing something... and i'll bet it's something inside qtractor's bare metal

hold on. now that you say you have a working rosegarden export file, how about doing some A/B (con)test? i'm asking for you to build two similar sessions in terms of tracks and instrument layout, one on qtractor and the other on rosegarden. then export to midi.

let us compare

nb. performance content don't need to be the exactly same, just having equivalent tracks/channels and their respective instrument bank/programs assignments each will be enough. that's the case in investigation

seeya

[Hoff123]

It's just that since (to me at least) the problem seems to be with Qtractor, and you being the creator of it and all... yeah .

Btw, I tried an exported MIDI file from Qtractor on an other PC, and yeah, still same problem...


And to what you said, sure. I'll do it . I'm not that familiar with Rosegarden but it didn't seem too complicated at least...



Update OK, so I have now made 2 "songs"(really, it's not even a song...) that are EXACTLY the same using first Qtractor and then Rosegarden. And I had some weird MIDI problems with Qtractor(the ones I mentioned, plus it also playing an ocarina instead of the normal piano for some weird reason outside Qtractor), while Rosegarden worked perfectly, but honestly I don't like it as much as Qtractor . Oh, and when I import an exported Qtractor MIDI file INTO Qtractor it works, but not if I play it via for example Rosegarden or Audacious(all using Fluidsynth). The exported MIDI file from Rosegarden works in all programs... And by "works" I mean they play the wrong programs/patches...

Btw, the patch numbers are 1 wrong depending on if you count from 1 or 0 lol...

Here are the files: http://www.mediafire.com/?6wisetj9cv9q531

[Hoff123]

rncbc, are you still alive?

[rncbc]

Btw, the patch numbers are 1 wrong depending on if you count from 1 or 0 lol...

is this the whole issue? a pretty old iirc. way down to primordial times of midi ...

like whether midi channel should be read as addresses (0..15) or ordinal numbers (1..16). but with program-change numbers it's a bit different and, well it's kind of a a dead end. by convention, some will get that program-change numbers are to be read in the 1..128 range when others stick with the bits of a 0..127 scientific truth or not. you see, it depends on the instrument definition whether it deos the [0..127] range or the [1..128] one.

well, qtractor, by default, sticks to the later. others probably make it in the former way and add +1 to the program-change value number anyway,... some others might even assume it's the other way around and then makes it by a -1 ...

i guess we've been just dwelling on old midi swampy waters all the time

cheers

[Hoff123]

Btw, the patch numbers are 1 wrong depending on if you count from 1 or 0 lol...

is this the whole issue? a pretty old iirc. way down to primordial times of midi ...

like whether midi channel should be read as addresses (0..15) or ordinal numbers (1..16). but with program-change numbers it's a bit different and, well it's kind of a a dead end. by convention, some will get that program-change numbers are to be read in the 1..128 range when others stick with the bits of a 0..127 scientific truth or not. you see, it depends on the instrument definition whether it deos the [0..127] range or the [1..128] one.

well, qtractor, by default, sticks to the later. others probably make it in the former way and add +1 to the program-change value number anyway,... some others might even assume it's the other way around and then makes it by a -1 ...

i guess we've been just dwelling on old midi swampy waters all the time

cheers

Huh? That's not the problem. I was just saying that the patches might be one wrong because of the counting, but it still doesn't mean it works... For example, one midi that was supposed to have a piano track was instead an acoustic guitar(like 30 other patches between them...) and some tracks didn't even have any patches at all...

[gordon72]

I think I have a solution for this:

If you bring up the note editor by double-clicking a track in qtractor. You can enter a 'program change' event by selecting 'program change' in the second drop-down. The pencil tool allows you to add the event. Drag the height of the event until you get the program number required.

There are some quirks for this:

Firstly, I find the number you want is probably one less than the program number you set in the qtractor track settings. No doubt because program change event 0 is the first in this scheme.

Secondly, and I am not certain about this, but I think it is more likely to work if the event is not at the very start of the track (ie time zero). Some sequencers seem to fail to register events at time zero.

This has worked for me when trying to create midi file exports from qtractor to be converted into wave files using fluidsynth from the command line. Without the 'program change' event in the tracks, the result was every track was 'grand piano'

[rncbc]

I think I have a solution for this:

If you bring up the note editor by double-clicking a track in qtractor. You can enter a 'program change' event by selecting 'program change' in the second drop-down. The pencil tool allows you to add the event. Drag the height of the event until you get the program number required.

There are some quirks for this:

Firstly, I find the number you want is probably one less than the program number you set in the qtractor track settings. No doubt because program change event 0 is the first in this scheme.

Secondly, and I am not certain about this, but I think it is more likely to work if the event is not at the very start of the track (ie time zero). Some sequencers seem to fail to register events at time zero.

This has worked for me when trying to create midi file exports from qtractor to be converted into wave files using fluidsynth from the command line. Without the 'program change' event in the tracks, the result was every track was 'grand piano'

the right way to do just that on qtractor is to set the instrument, bank and/or program at the track properties level--that will make it to a correct program-change at the beginning of the exported sequence.

although interleaving a program-change within the midi sequence is perfectly legal, it is NOT as reliable as the standard convention for which qtractor sequencing model follows straight: simply said, one track, one channel, one instrument, one patch (bank_select + program_change).

hth.
cheers