Wrong math?

Topic: Wrong math? (Read 8541 times) previous topic - next topic

Wrong math?

16 years ago

Look... well, listen to that F# in bar 3.
Why it's so long?
Did I the math wrong?

Re: Wrong math?

Reply #1 – 16 years ago

The quarter = 40 tempo comes in at the third beat where you have it. The f# comes in half way though the second beat.

Try putting the tempo marking on staff 3 before the f#. That will fix the problem I think.

Re: Wrong math?

Reply #2 – 16 years ago

No, Richard, it doesn't work.
Even removing both the tempo changes (20 and 40) the F# is still longer.

Re: Wrong math?

Reply #3 – 16 years ago

Standard NoteWorthy treatment of non-legato notes of differing duration. If you want all the notes to end at the same time, put a tenuto on them. Or put them all on the same channel and use Pedal down/release.

Re: Wrong math?

Reply #4 – 16 years ago

Good old Rick on the rescue: you got it!

I suppose this has to do with the normal note shortening inserted to not always have the legato, but I don't understand why notes of different duration should end in different instants.
Are the small gaps proportional to note duration?

Could you please explain it with more detail?

Re: Wrong math?

Reply #5 – 16 years ago

See the attachment. It has nearly all the duration and articulation combinations that NWC can produce from untied notes. The top lyric line is the # of clicks (Parts per Quarter Note) from note on to note off. The second lyric line shows what % of the notes total duration that it is on. For tenuto notes, this is 100%. There are some large and unexpected transitions here, which may help explain why notes that look like they should end together do not.

Enjoy

Re: Wrong math?

Reply #6 – 16 years ago

Interesting.

Quote

Calculated by sending this file to midi and analyzing it with a script.

You mean you exported it to MIDI or you captured the MIDI output of NWC?
I think to remember that there can be some difference.

Quote

There are some large and unexpected transitions here

I'm surprised.
When you have to divide, let's say, 10 ticks on three notes then I understand some jitter and you need to find a solution of compromise. But when workiing on hundreds of tics I don't understand why such a big spread of ratios.

Anyway, I think it's the fact of having the gaps proportional to note duration that can lead to the effect I encountered.

Thank you Rick.

Re: Wrong math?

Reply #7 – 16 years ago

Quote from: Flurmy – 16 years ago

You mean you exported it to MIDI or you captured the MIDI output of NWC?
I think to remember that there can be some difference.

I exported it to MIDI. Fermatas and Breath marks with delay are different in NWC playback vs. MIDI export, but there are none here. If I captured the realtime output, I'd end up measuring the latency of MIDI-OX and possibly my software synth. I finally retired my SB AWE32 (and its hardware synth) after 10 years. ISA cards and WinXP don't like each other very much

I also had to retire my 17'' CRT monitor. Win98 does not like my new fixed pixel TFT display. There weren't many 16:9 monitors around in 1998.

Quote from: Flurmy – 16 years ago

I don't understand why such a big spread of ratios

I see 2 problems:

The linear functions should be exponential
Augmentation dots need to figure into the calculations

The problem with changing it now is that it would affect playback of every song that has ever been done.
NWC was not intended to be a sequencer. That it has turned out to be as good at sequencing as it is, surprises me. It is good enough for GM1. If the decision were made to fully support GM2 and DLS, it would need a major overhaul.

Re: Wrong math?

Reply #8 – 16 years ago

Quote

If I captured the realtime output, I'd end up measuring the latency of MIDI-OX

If the latency of MIDI-OX is constant, as it must be to have a jitterless playback for MIDI is a real time interface... or at least it should be, then all you have is a simple time shift (delay) that's completely unimportant in this case. No other instrument to be in synch with and no reference to any "absolute" time.

Quote

and possibly my software synth.

I really can't get what a SW synth has to do with the capture of a MIDI event stream.
Anyway this discussion is purely academic. Dont' worry.

Quote

I finally retired my SB AWE32 (and its hardware synth) after 10 years...

I retired my new computer instead.
Heck, I mean, the motherboard retired itself from work...
N.B. My 486 vintage '92 keeps working fine. ;-)

Quote

* The linear functions should be exponential

Which functions do you mean?

Quote

* Augmentation dots need to figure into the calculations

Uh? Do they not?
The percents are almost the same with and without the dots: 100% - 93% - 62% - 25%

Re: Wrong math?

Reply #9 – 16 years ago

Quote from: Flurmy – 16 years ago

I really can't get what a SW synth has to do with the capture of a MIDI event stream.

I'm not an expert on how Windows handles MIDI but, it wouldn't surprise me if when it talks to a software synth, it uses some handshaking. This would make the capture dependent on the software synth's ability to handle the stream.

I think that using the MIDI file gives the best indication of what NWC2 is doing, independent of all the subsystems in the chain that are needed to convert MIDI to an audio format.

The linear function I'm referring to is that from measure 9 onward, the difference between Tenuto and Semplice is 2 PPQN's. Surely that is too simplistic.

My comment about augmentation dots stems from the big transition between measures 8 and 9-10. For semplice notes, there is a marked difference between quarter and '(Dbl)Dotted 8th'. (83% vs 99%)