Based on a quick bit of Google-ing, I note that this device has a built-in MIDI record facility. I assume therefore that you spot the doubled notes on the recording made there? So i'd guess that it's not because of some problem with cables, or a computer, etc.
It would be interesting to see the midi file created that includes this.
I assume that the note that is ON can sound once only, and that you do not hear anything amiss?
I'd wonder if there is something very slightly wrong with one or more of the keys, regarding the point at which the ON is triggered, and the OFF is triggered, such that on occasions you might play in a way that just allows the triggering of the first ON, the slight release of the key that does not in fact trigger the OFF, but still allows a later trigger of another ON?
Does this happen with certain keys ONLY, are can you achieve the same effect with any/all keys?
It's fairly large, and there's a LOT of data there, all on the same channel.
If you've spotted some problems there, please note the approx note and tick cound where these show.
I was hoping to see where the problems occur relative to use of controllers, and what the parameters of the 'odd' notes are relating to others before/after. i.e. is the problem happeninf with short or fast notes, or loud notes followed by soft ones?
I DO see that your piano is generating a LOT of controller settings re Hold pedal, so there can be a LOT of controllers to handle in a VERY short space of tim,e, and this may occasionally swamp the system? But this is just a guess right now.
Technically, if you encounter a note starting message for a channel and pitch that is already on, it means a new instance of that pitch has started.
The problem is that once there are multiple instances of the same channel and pitch that are on, when you encounter a note ending message for that channel and pitch, there is no standard way to tell which note starting message should match up with the note ending message.
Because of this, MIDI devices and software usually avoid having a note starting message for a channel and pitch that is already on. If you encounter this, it is likely an unwanted glitch.
In the MIDI file you attached, there are three instances where a note starting message appears for a channel and pitch that is already on, see the notes highlighted black in the attached piano roll pictures. I can only guess this is the result of some glitch, the resulting overlapping notes likely should have just been one combined note.
OK, thanks Bavi - your images helped me find the actual midi data involved here.
Firstly, the octave numbering between Bavi's system and my data seems to be different. Bavi's data shows notes E4 and A3, in my data it seems to be E3 and A2.
That aside, I see what seems to be the same problems. Rather, oddities.
The location numbers are the midi tick counts in the raw data.
For the first case:
at 13680 I see a Note On for E3
almost immed there are damper events at 13681 and 13682
at 13696 I see another note On for E3
at the same tick there is a Note Off for E3
at 13707 there is another Note Off for E3
I suspect that this OUGHT to be a Note On between 13680 and 13707, but the Damper has caused a problem and a spurious On/Off in the middle.
I would guess that the On/Off would not be heard.
For the second case:
at 18886 there is a Note On for A2
18932 Note On for A2
18932 Note Off for A2
18960 Note Off for A2
No Damper events here, nothing else odd. The spurious On/Off is not explained.
at 23729 there is a Note On for E3 velocity=47
at 23741 there is a Note On for E3 velocity=15 (which is unusually low ref other notes in the same area)
at 23741 there is an Off for E3, showing a velocity=0 (which is odd as the Note Off are explicit Off rather than Note On with Velocity=0)
at 23807 there is a Note Off for E3, velocity=65
Here, the second On is wierd, and the immed Off is also wierd.
In the first case, the Damper MIGHT have interfered with the data, but the pair of data On/Off was in fact there. The data I'm looking at does not show any missing events, it's just the order of the events that is messed up.
In all three cases, there is, in effect, a spurious On and an immed spurious Off, and in all three cases this spurious note might not be heard at all.
BUT, this depends on the playing hardware, and how the system would handle two notes the same sounding at the same time. As is suggested above, you cannot be sure that any system would see such cases as two separate notes, with overlapping On/Off, or would just end the note at the first Off and disregard the second Off as the note was not in fact On any more.
I wish I knew what condition causes my piano to generate these extraneous MIDI events, but if every such ON is "followed" (with 0 delta time) by an "OFF" and each such ON-OFF pair is actually meaningless, then I can ignore all such ON-OFF pairs, whether or not I am simulating the dampers being up because of the sustain pedal.
In which case my real problem with interpreting the MIDI file the piano created is in deciding what note ON events are meant to be inaudible because their velocities are too low.
(I am using the MIDI file to create sheet music, but I don't know any music notation for a note's "velocity", let alone zero length.)