Chord sequencer export to midi then import again - broken?

[unjazz]

I've been tearing my hair out all day trying to figure out why exporting from the STYLE PLAY chord sequencer, then immediately re-importing that midi file into the chord sequencer again does not give me back my original chord progression. Is something broken on the PA700 or am I doing something wrong?

if I save and reload the chord progression via a midi file - I DO get most of my original chord progression but minus the last chord or so - resulting in whacky timing. E.g. If I record a 8 measure chord progression (e.g. style "Let her Ballad") of

Code: Select all

C D E F G A B Am

it plays back as expected. viz.

Code: Select all

measure   chord
1  C
2  D
3  E
4  F
5  G
6  A
7  B
8  Am
...looping now...
1  C
2  D
3  E
4  F
5  G
6  A
7  B
8  Am

Saving the above style containing the chord progression and reloading the style plays back the chord progression as expected. So far so good.

HOWEVER

If I export the chord progression as a MIDI file, then import it (bars one to eight) back into the style chord progression, playing back the chord progression gives me

Code: Select all

measure   chord
1 C
2 D
3 E
4 F
5 G
6 A
7 B
8 C <--where is the Am?
1 D
2 E
3 F
4 G
5 A
6 B
7 C <-- again, have skipped Am and gone to C
8 D

See how the last chord of the chord progression is missing and the timing is out of whack? I've tried numerous styles and export/import combinations incl. adjusting the number of bars being imported back in, but cannot seem to re-create the original chord progression.

IMO this is a serious bug for me. Not being an expert chord player, I have complex jazz chord progressions that I STEP enter in the Korg PA700 sequencer or on my computer (from books etc), and cannot reliably import them into chord progressions!

[karmathanever]

Hi

I just copied your example exactly on my PA4X and the export and import worked perfectly.

I don't own a PA700 but perhaps someone else could try this for you...

Pete

[unjazz]

Hmm thanks for trying that out. I ocassionally get an OK export/import cycle but 95% of the time the last chord is missing after importing. Perhaps you were lucky, or the PA700 is slightly different? Or perhaps you are recording chord progressions in a more accurate way than me (I hold down that last chord for as much of the last measure as possible, before hitting the "chord progression" button or STOP/PLAY to stop recording the progression, which I hit just before the first beat of the next measure, thus avoiding recording into the next variation cycle).

My findings:
By examining the exported midi file (via Logic Pro X or via Korg PA700 Song Play / Score view) it turns out that the midi export of chord progressions is hit and miss (mainly miss) re how long the last chord of the progression lasts. Even if you hold down that last chord for e.g. four beats of a 4/4 measure, before hitting the "chord progression" button or STOP/PLAY to stop recording the progression, the last chord may be recorded and exported to midi as merely 1 beat long?!

E.g. https://www.dropbox.com/s/q4en3j5jp6msv ... .png?raw=1
Image of a chord progression exported Midi. The last chord is exported as only one beat of the measure long - despite being held down for the full measure, thus causing the midi import problem.

Whilst the original chord progression one records on the PA700 plays back and loops ok in STYLE PLAY mode, after exporting the progression to midi and re-importing it, that last chord may not be long enough in duration ? to be recognised during import - resulting in the chord progression of the last chord of the midi file to be skipped (not imported). This of course is bad and also throws the whole loop of the chord progression off as we loop.

A Solution:
The only fix for this seems to be to edit the exported midi file (via computer or perhaps using the Korg PA700 sequencer) to fix the length of the last chord to be longer and e.g. fill the bar. I haven't explored a fix for other chord progressions where one might actually want to have a lot of shorter one beat chords at the end of the progression.

P.S. Then there is the issue of the timing of the chord progression changes - the chord transitions often feel late - the old chord momentarily sounds on the beat, followed a few milliseconds later with the correct, new chord - creating a jarring playback. E.g. on style "Years Ballad". But I suspect this is another topic...

[unjazz]

Actually playing around with chord progressions some more e.g. "Fast Jazz Waltz", there seems to be a black art to getting them to record correctly in the first place (let alone the midi import/export issue).

Recording a progression, keeping in my mind the variation length of the style, and hitting STOP/PLAY at the last possible moment of the last beat of the last measure of the variation cycle
- Usually gives me a clean loop maybe 90% of the time if its a one variation cycle e.g. 8 measures (assuming "Fast Jazz Waltz").
- If I try a two variation length cycle (16 measures, assuming "Fast Jazz Waltz") chances of nailing it drop? though see update below in next post where I had much better results .

Example of a chord progression attempt:
I used to use sync start, but now I do this technique hoping for greater accuracy:

Code: Select all

- arm the chord progression - red light comes on
- play Em
- hit STOP/PLAY... off we go... red light blinking
- wait 4 measures as the Em is recorded
- Am+9 for 4 measures which nicely fills up the first turn through the variation
- CM9 for 2 measures
- Bm+9 for 2 measures
- CM9 for 2 measures
- Am+9 for 2 measures (thats four x two = 8, the variation length)
- hit STOP/PLAY at the last possible moment (as described above).

Play it back, and in this attempt, everything is perfect except I get the last Am+9 for three measures instead of two, thus throwing the whole loop off?

[unjazz]

Update: After dinner and a break, I tried recording chord progressions again and they all work. In var1, both one variation length (e.g. 8 bars) and two variations worth (e.g. 16 bars) and the loops are OK. I tried both synchro start and non synchro start techniques, across 9 styles - they all recorded correctly.

I don't know what is going on - at least its some good news. I swear I was getting anomalous results all day which inspired my previous post. I'll try again tomorrow to verify. At least (unfortunately) the midi export/import problem persists and is easily reproducible (for me!).

[karmathanever]

That all sounds very frustrating - I do not experience this on my PA4X - yes I do get that "extra" measure on the end but that is purely my fault.
I assume you have the very latest OS update etc.... (just checking)

I really hope another PA7600 owner can do some checking for you...

Cheers

[unjazz]

> yes I do get that "extra" measure on the end but that is purely my fault.

Nice to know! Agreed, Ive figured out that the reason I'm getting better results recording chord progressions now is that I've unconsciously improved my technique. Which I think is: for recording precise chord progressions that record the last chord correctly and loop correctly, you need to hit stop ON the first beat of the next bar, in time. Not before the end of the last bar. Recording similar loops in e.g. Ableton Live software is a bit more forgiving and "rounds things off", the Korg is not forgiving.

More clues re the chord progression import export problem:

Here are two chord progressions that loop perfectly well within a style, and are exported as midi. When imported into a chord progression, the PA700 doesn't import the last chord. Hopefully PA users can try importing these two midi files and report what they get.
https://www.dropbox.com/s/zh6hex6pkqpb1 ... .zip?raw=1

I've also just discovered that if you load these midi files into the Korg SONG PLAY and play them, the last chord doesn't even play on the Korg! Look in score mode and you can see the chord (albiet shortened) is there, but it does not play. No wonder it that last chord doesn't import into a chord progression. So I think I've narrowed it down to an error in the way Korg chord progressions are written out to midi 1) last chord is too short in duration, 2) last chord doesn't play. Loading the midi files into Logic Pro - and the last chord, yes whilst short (should be the whole measure in duration) at least plays OK. But not on the Korg.

[unjazz]

I've generated a Youtube video illustrating how to easily reproduce this problem.
https://youtu.be/bsYkj7Mu6yo

[siebenhirter]

... I do get that "extra" measure on the end but that is purely my fault. ...

I think Midi-Chords of your Pa-arranger generate defect midis. If you load them into sequencer of any Pa-keyboard the last chord notes will show you a length of zero.
*
I used the SMFs downloaded from you because could not test this with one of the recent Pa-arrangers like Pa700/1000.

Started some DOS procedures (MIDIFIX (c) 1995 created by Guenter Nagler ) and got following message with ChordSeqEg1A-B-G-A.mid:
"16 byte garbage at end of track 1" - length of track 1 updated
remove garbage at end of midi (16 bytes)
.mid contains errors. Fixed output written to ...
mthd
version 0 // single multichanneltrack
// 1 track
unit 384 // is 1/4
tact 4 / 4 24 8
end mthd
Track 1 [1]: "" (GrandPno)
Channels: 1
Notes: d2 g2 a2 a#2 d4 f4 g4 a4 a#4 c#5 d5 e5 f5 g5 --> but that SMF is shortened with the last zero length (D:0.0.0) chords notes
Velocity: 100
... etc
*
A (provisional) solution is to load the SMF with an external sequencer and then save it as an SMF0. Loading such rewritten SMFs with internal sequencer (also of older Pa-arrangers) has no faulty lengths of lost chord notes and does not cause an error.

Such SMFs corrected in this way should therefore be suitable for re-import as chord-progression files, in which the last chord notes no longer are missing.

I tried Cakewalk (Pro9), Mixmaster (freeware), PsrUti (freeware) and each of them will give you an SMF with the correct lengths of the last chord notes.

However, there is some text in the SMF from Mixmaster and PsrUti (Marker Edited by Mixmaster / Psruti ...), but I think each other sequencer software that is not freeware will do the same job.