Originally Posted By: Mike Head
Hi
I have followed this thread with interest.
But I think at the end of the day you will have to except that Midi and Notation are two very different forms of music data.
As has been said midi is a series of commands to play a voice be it trumpet or piano and contains amongst others Note on Note off = note length, and Note Velocity in steps 0-127 = how loud the note is (this is done by measuring how long it takes for the key to descend the harder you hit em the faster they fall)
How hard you have to hit the note on a keyboard is dependent on your touch setting.
These velocity values can also be input manually in step mode in a DAW .
Midi can not read hairpins and expression from a score or slurs etc.
It really is meant for playing the notes in from a midi device, be it wind controller or keyboard.

Have a look at this below:
Coppied from, MIDI tutorials by SysExJohn

, an extremely good idea if we could map these concepts of loudness to MIDI note velocities. Lo and behold it has already been done. In XGworks, the first "proper" sequencing program I ever used, it uses ppp through fff against the note velocities in both the "staff view" and the "piano roll" windows and also within "list view" window too.

It works like this:
fff = 120.
ff = 104.
f = 88.
mf = 72.
mp = 56.
p = 40.
pp = 24.
ppp = 8.

These are, in fact, centre values, so for instance ppp is from 1 through 8 to 15, pp is 16 through 24 to 31, and so on.

How does this work in practice? Well, using my MU1000 sound module and measuring a note played at the centre values I found a difference from loudest to softest of approximately 52 decibels. Quite a significant dynamic range. In fact with CC#7 set to 100 (MIDI default) I found it difficult to hear the softest note (ppp = 8). Changing CC#7 to 120 made it audible, but then the loudest note (fff = 120) became almost painfully loud!

But you still have to play them in or enter them in step record mode .
I do not know of a score reader midi player that can do this.
Still maybe the values will be of help.

Mike


















=





Hi!

You guys are so darned smart! This issue presented here has been going on for a few months now, off and on, came to a head with me posting here, as thorough web searches as I could do answering nothing explicit enough to trust. Again, nobody on the planet, it seemed, could orient me with the plain facts of the matter you guys have posted. As a matter of fact, all the "expert" advice was always you have to buy super duper sound libraries. I would only have wasted money on pricey products I don't own, turns out wouldn't use, at least for non-orchestral, and even controller mechanisms lacking in that instruments array, albeit not all generally atrocious like piano playback. Clearly, many people, it seems most everybody, even using notation products, are not doing notation, at least the old fashioned way, rather using midi controller inputs. It's another discussion why I don't do this that goes to discipline and premeditation, thoroughness, but it sort of blows my mind nobody was able to flatout tell me a midi notation export has insufficient data to sound like the real thing. You can say it in one sentence, IF you know this, and follow that with don’t waste your money, hope you live long enough to see a truly AI intelligent product. Incredible to me the pan-with-flanged-echo-and-4-FXs-on guys, including ostensible notation experts, would say stuff like "you need Miroslav" or East West. Incredible notation websites don't address this issue, at least that I could find in a reasonable amount of time looking. Of course, the notation products companies aren’t likely to advertise notation playback midi sucks, rather give you a much tweaked demo that sounds like Carnegie Hall. I chased my tail for a long time, looking for some evidence a couple expensive libraries would do what I wanted to do, which, turns out, isn't possible, sans major manual processing of notes. I can only conclude people are automating and most not actually writing music, at all, as was done the past hundreds of years.

Anyway, thank you much for the values! I did see something in a YouTube video where a gentleman was setting velocities better, but couldn't see what he did on the screen. That is excellent information. You guys really know what you're talking about, and it would have saved me such wasted time and effort to have posted here, first, as, in a merely a couple pages and same day, you've hashed-out the matter. And I hear you, no pun intended, about those sudden-on dynamics, if a hairpin isn't the best notation leading there. I will proceed to at least set some better velocities, as forte+ wants to tear the speaker cone. (Leave it up to notation playback, you'd name all your works "Alarm Clock 1", then 2, 3...) Thank you again!