Skip to main content ↑ ↓
Topic: Ghost note (Read 4145 times) previous topic - next topic

Ghost note

Where does that last top F# comes from?
A can't see any reason!

Re: Ghost note

Reply #1 –
Hey Flurmy,
Quote from: Flurmy –
Where does that last top F# comes from?
A can't see any reason!
I must be missing something...  I've downloaded the file, opened it in NWC it plays as it reads.
Checked the nwctxt and it is exactly what I would expect.
From the key sig F's are sharp anyway, but the ties make each chord the same anyhow.
I plays 'Bones, crumpets, coronets, floosgals, youfonymums 'n tubies.

Re: Ghost note

Reply #2 –
I don't understand.
On my computer there's a new key on at the last top F (F# indeed) and only that note, even if the note is tied.

I tried it on my daughter's computer and everything is ok (windows 10 & VirtualMIDISynth).
I exported it to MIDI and it's ok also on my computer.
If I shif it down one octave it's ok.
If I shif it down one fifth it's ok.
If I shif it up one second it's ok.
If I shif it down one second it's ok.

So, who's to blame?
NWC?
VirtualMIDISynth (an old version: 2.9.3, the last usable on XP)?
Windows XP?
N.B. Of course, the MIDI player uses the same VirtualMIDISynth as NWC.

I'm completely clueless.

Re: Ghost note

Reply #3 –
If I change the soundfont the problem disappears...  :o

Re: Ghost note

Reply #4 –
One possible effect is that if some notes in a chord fade out faster than others, we (or at least I) at times seems to "hear" the slower-fading notes anew. I do many arrangements for crank organs, where I use "Ocarina" of the Chorium soundfont as a sound that is very near to that of smaller crank organs - and I have had such effects at times, which made me search for spurious notes that simply were not there. (In one of my first arrangements, I found out that at least on faster scales, you can trick the ear into "hearing" notes that are definitely missing ... we seem to just "know" what "should be heard", and then hear it ...). This is not a definitive explanation, but I am not totally surprised about this.

H.M.

Re: Ghost note

Reply #5 –
I understand what you describe since I already met it but, sorry, no psychoacustics this time.
Just plain old hearing!  :D
Listen yourself.

Go figure...

Re: Ghost note

Reply #6 –
Nice :-). Which soundfont?

I overlayed your Ghost.mp3 with a MIDI recording of the NWC file - the Reaper result can be seen in the attachment: This is really very weird: The ghost f# occurs "somewhere in" the 6th beat! (I hope the tempo maps are actually the same; but why shouldn't they be?) ... scratching my head slowly...

Re: Ghost note

Reply #7 –
I had little doubt the problem is caused by the "note off" event who, I don't know why, becomes a "note on".
As now I expected, even driving an external MIDI keyboard from NWC is ok.
N.B. The soundfont is Airfont 340.

Re: Ghost note

Reply #8 –
Hey Flurmy, pretty sure it's the soundfont.  And it seems to only be the Acoustic Grand Piano patch too.

I'm also using VirtualMIDISynth and didn't have a problem (my current soundfont is Merlin_GMPro(v3.15).  I didn't have a copy of Airfont340, but I did have a copy of Airfont380_Final and it was also fine, so I hunted around and found a copy of Airfont340 and voila, there it was, but as mentioned, apparently only patch 0 (the Acoustic Grand).  NB I didn't check all the patches, just a small sample.

Actually, to be more correct, it's possible it's the combination of the soundfont and Coolsoft VirtualMIDISynth and NWC.  To be certain a different synth using the airfont340 soundfont should be tested. 

NB IIRC NWC uses a Note On with velocity zero or the like as a note off.  Don't remember exactly, there was some discussion on the forum about this relatively recently.
<edit> found it.  You even commented ;)
https://forum.noteworthycomposer.com/?topic=10064.msg72304#msg72304
</edit>

Maybe just checkout Airfont380 and see if you like it...
I plays 'Bones, crumpets, coronets, floosgals, youfonymums 'n tubies.

Re: Ghost note

Reply #9 –
I tested Airfont380_Final and the result is: no problem.
I can hope the bug was fixed, but remember that the problem appears only in a very particular case.
Who knows if, by chance, the problem now appears only for, say, Bb2?