KSEQ to ESEQ file conversion

[bnz99]

By the way, I'm saving a bunch of SY77 sequences that I know pretty well that I can test playback from Cubase via Midi out to to the SY77.. This should be a pretty good way to test if the songs sound right.

[animaux]

Great, thanks! You have just created the longest possible note on a SY sequencer

Code: Select all

11011111 01111111

[animaux]

By the way, I'm saving a bunch of SY77 sequences that I know pretty well that I can test playback from Cubase via Midi out to to the SY77.. This should be a pretty good way to test if the songs sound right.

Great, looking forward to your feedback!

[bnz99]

Is it already fully updated? My impression is that it's quite a bit better. But some tracks are still off with their starting points. Oddly, in the one song I have tested, only 3 tracks are misaligned. The remaining ones are fine. All these tracks have in common that they start very late in the song. Apprximately: one track had to be shifted roughly 64 bars, one track about 45.something and one track about 51.something bars to be roughly in the correct spot. The numbers aren't exact as they are all misaligned differently. Two of these misaligned tracks start (wrongly) at bar 1. One starts at bar 2.

Should I maybe create a test track with:
1) track 1 a c note full bar at measure 01.
2) track 2 a c note full bar at measure 46

If that converts with the same wrong starting position, you'd have a pretty minimal sample.

[bnz99]

There is a new file DEL.K02 in the dropbox folder. it's exactly created as written above. The midi file however starts track 2 on bar 2 instead of bar 46.

[animaux]

Oh damn, it seemed like everything was fine in all the sequences I’ve tested. But here we are, I’ll have a look

[animaux]

(I get a whole bunch of notes on a Track 16 listed in the analyse part, but nothing of that is showing up in MIDI file. Strange. Actually Track 16 is reserved for pattern from pattern mode, at least the manual states that. Maybe this has nothing to do with the current offset problem, though.)

OK, looks like the ticks are calculated right, so there is probably an error in my MIDI delta constructor. First note of Track 02 should start at 17280 ticks /96/4 = 45 bars.

UPDATE found and fixed the error. Your file was perfectly helpful, thanks again! Actually the line for creating 3-byte MIDI-deltas was commented out …

[animaux]

We could do a final delta-test if you want, to see if four bit deltas work as well in MIDI. We would need the same file, but with the second track object at least at 5461.333 bars. But we don’t want to do that, or do we?

The limit for deltas is 4 bytes, which would amount to 699050.666 bars (@96 ticks per quarternote). I guess they decided it’s probably more than anyone will ever need, except John Cage. But then you could reduce the tick resolution or play the song slower, so even he would have been content.

[bnz99]

The first test was flawless. Check the dropbox folder for the mp4 file. I have recorded the Cubase playback. Checking the uncle poe demo song next.

[animaux]

Very happy me, thank you so much!

[bnz99]

The second test with uncle poe was also flawless. Very cool! mp4 is also uploaded. I'll do one more test for today.

[bnz99]

The stepstyl demo song has a problem with track 16. The odd thing is, everything is fine if the track just didn't play at all. I'll do some inspection on that tomorrow, not sure what is going on with track 16 in the original. Possibly some missing muting or velocity off functionality? If have added the mp4 and the stepstyl.k05 to the folder.

But absolutely awesome work so far! Who would have expected that this converter would still happen

[animaux]

\0/

I didn’t think this was possible too, only two or three weeks ago.

Not sure about Track 16 either. The manual says in Song Mode you can use all 16 Tracks, but then it also says Track 16 is exclusive in Pattern Mode. I have only one sequence where I’m sure pattern mode is used. But that one doesn’t even use Track 16. So I’m not sure what to make of it.

There’s a lot of further processing needed to make conversion of KSEQs with pattern mode possible. If I understand it right the patterns are laid down similar to the normal track data. But the pattern allocation table stores their exact position and the order for playback. At the moment you should get all the parts of a Pattern Mode KSEQ, but obviously not in the right order, count and position. Not sure I want to venture into this too.

Your DEL.K02 file did have a lot of notes in Track 16 too. But in its case »Tracks used« only shows «1,2«. So there is probably some date stored for Track 16 in the file, but Track 16 is disabled? But it’s strange that its not showing up in the MIDI file, because the converter doesn’t take the »Tracks used« entry into account when scanning the actual track data.

Hmm. Looks like in MIDI the chunklen for Track 16 is not calculated correctly. MIDI File looks right, has all three tracks and the number of tracks is also stated correctly.

[bnz99]

I'll do some tests in the evening to figure out whats up with track 16.

[bnz99]

I have checked this. It's like this: the stepstyl song indeed has two patterns with each 4 bars.What is played in the beginning of the midi conversion is the 4 bars of the first pattern. The second pattern doesn't seem to be part of the midi conversion. According to the manual, it states the following about track 16:

- p. 9: ... arrange rhythm patterns in track 16 (the pattern track)
- p. 16: ... You can create 99 patterns and place them in track 16 (the pattern track).
- p. 17: Track 16 is a dedicated Pattern track. In pattern mode, you can create up to 99 patterns; short phrases on to thirty-two measures long. Patterns are especially suitable for rhythm parts;...
- p. 35: Each track 1-15 contains an independent musical performance; notes, pitch bends, controller movement, and program changes. Track 16 is a dedicated Pattern track. It contains pattern numbers and repeat marks. When playback comes to a pattern number, the specified pattern will be played.

Screenshot 2024-11-28 135916.png (134.46 KiB) Viewed 70 times

- p. 38: WHen you enter pattern recording mode, the keyboard will transmit the channel selected for track 16 (the pattern track) of the sequencer.

I still don't enirely undestand how this works. I'll have to try this out more practically. But nevertheless, I would think that you'd either have the choice to support patterns, then more work would be needed or you choose to avoid pattern. In that case, I would put maybe a pointer on your website that patterns are unsupported and that track 16 may contain incomplete patterns. Personally, I have never used patterns back when I used the sequencer. I'm guessing most other haven't as well.