I have not used midi files much but am thinking about playing around with this concept. I am wondering what ways would be best to get them to sound their best on my Tyros 2.

One thing I have noticed about midi files is that there seems to sometimes be a quantize issue, for example if a piano player does a trill or a run of very fast notes sometimes it just doesn't sound right, it sounds like a dumb machine Is there a way to correct this?

What about using the best instruments to play on the midi. What about other effects that one might want to consider when tweaking these?

Are you using software programs on the computer to do this or do you use the onboard gizmos on your keyboard.

What about normalizing the volumes between songs? We tried Michael Bedesem's program for this once, and I don't know if my wife didn't do it right or if this is not all that accurate of a program.

Thanx
Scott

Hi Scott,

On the first issue there can be a couple of reasons. One is that the notes were "pre quantized" or laid out so that they stuck tightly to a grid, sometimes to the closest 8th or 16th note division. If they were step recorded, it's also likely that all the velocities ( note strike pressure ) are identical. Or they may have been recorded live and then were quantized. Another problem can be that they are recorded in sequencers with low resolution.. .ie 96 ppq for example. Any of these things will definitely contribute toward that "mechanical" sound that we don't want. Some sequencers have "humanization" functions, that will allow the user to randomly move the notes slightly and / or change the velocity. These work to varying degrees.

Usually when tweaking voices, I focus on changing the actual voice itself, then the reverb and chorus effects. Some instrument sounds will need more than that.. overdrive guitars for example, usually sound flat or "synthlike" without some overdrive effect applied.

If I am tweaking a midifile for use in a hardware board, I probably use the board's sequencer more than a software sequencer. The exception being when I want to edit notes to get rid of the stiff sound or quantize effect. I don't generally care for the humanization features that much on my sequencers, and find it a lot easier to edit note times and velocities on a piano roll grd. Occasionally some of the cal scripts I've used in Sonar work out ok as some of these can also automate random properties.

As far as normalizing the volumes, like everything else in the midi realm, I find that the midi "standard" isn't so standard after all. That's why, for example, reverb level 40 on a drum track in my MZ2000 might be just the right amount for a certain midifile, while the same level on the same midifile played in the Motif ES might make it sound like the drummer was playing in a canyon or tunnel. Unfortunately, it works this way with just about all of the midi parameters you can think of, including volumes, velocities, etc. Noone uses quite the same samples or effects, so it really isn't surprising when you think about it that way.

We had this debate a few posts ago.. and there were those that suggested laziness on my part for not wanting to tweak all of my midifiles. In fact I've probably tweaked ( not converted ) more styles and midifiles than anyone I know of, and at this point I find it easier for me to use something like SGM180, which was made available freely by someone who was able to come up some great samples but also great compromises that make most of my unmodified midfiles sound their best.

Or to put it another way, although my Motif ES probably has the best sounds and certainly the most wav rom of my 3 boards, it is also the worst of the three for midifile playback. That isn't really even subjective, because it is so bad compared to the other two that everyone else who has heard it pretty much agreed. Yep, I can ( and have ) made some of my midifiles sound really good with it. The amount of work that was involved to do this wasn't worth the effort however, particularly when after all taht effort, something like SGM180 still sounds nearly as good right "out of the software box". The other thing is.. if I want to use midi files to do a job, and my ES suddenly goes "South", the midfiles modified for the ES won't sound any better if I need to use my MZ2000 as a tone generator, and it's likely in some cases they'll sound worse.

AJ

[This message has been edited by Bluezplayer (edited 06-04-2006).]

Hi AJ,
it was actually that post that got me thinking. I've got about 300 to 400 song files that playback in my Clavinova , but can't be edited. Most sound ok but the xg voices in the Clav are pretty weak.
I decided I'd edit the soundsource instead of trying to remap the program changes & drums in each of the songs.
Started editing the drums & sounds in the SD1,then realized, it's just as likely not to be around in a year or two. One of the companies will bring out the next new beaut arranger keyboard ( another must have) and all the work I've done for the songs will be wasted.
Instead I've now started editing a soundfont to match the program changes & drum mapping of these songs & with a bit of luck I won't have to worry about them again,.

best wishes
Rikki

Sorry Scott, I've gone a bit off topic.
[QUOTE]Originally posted by Bluezplayer:
[B]
We had this debate a few posts ago.. and there were those that suggested laziness on my part for not wanting to tweak all of my midifiles. In fact I've probably tweaked ( not converted ) more styles and midifiles than anyone I know of, and at this point I find it easier for me to use something like SGM180, which was made available freely by someone who was able to come up some great samples but also great compromises that make most of my unmodified midfiles sound their best.

Scott: here I wrote to Rikki the things I use/do when I change keyboards and re-map Midi files voicing them to take advantage of the BETTER sounds on the new unit rather than the 128 GM/XG ok sounds.

"Hi Rikki:
I did owned not just 2 SD1s, also a X1 too, well, the midi file re-voiced in song setup in the SD1+ has only the instructions of those changes only for the Ketron, but the midi file remains unchanged unless you remaped the voices and controllers in the PC with any secuencer program such like Cake/Sonar, etc....I did changed in the PC (with Cake/Sonar)the few midi files I did use, I also find that I have more control even in the dynamics, fx, etc...
As an example for my own experince is:
in the Reverb (control 91) to have a decent reverb in the SD1 you must set up at least 100-127 value(from 0-127), but in the case of the Yamaha Arrangers like the PSR3000, T2, you need less, around 50-70 depending on how much FX you like, each manufacturer FX procesor has diferent sensibilities, also may I point that also the note velocity response is diferent, I have to lower it if I play a file in the Yamaha, somewhere 50-80 note velocity range is ok, more, sounds overdriving it, but velocity ranges between 100-127 are just fine for the Ketron.
Needless to say, I do recomend to make the adjusments to each midi file in each track in the PC, well, at least I do it like that, perhaps because I am very picky.

In the drum maps, I still use the PC, because sometimes you get a file that has an annoying clap that should be a snare (as an example), that, only can be remap in the PC with the Piano roller mode, is a very fast procedure, this is not posible to do in the Tyros2-SD1+ yet, only change the drum kit and volume, fx overall.

Or you get a cymbal that is a little to loud, or the hihats to low, well, I use the Cals (Bustdrum) for Cakewalk/Sonar, so, I make the note velocity adjustment and that is it."

By the way, I do like to use Cake/Sonar software because you can add the "Instrument Definitions" for your particular brand/model, and that way you can edit the patches seen the real names of the sound banks, rather than a Bank number and a patch number, is easier and faster.
Also the CALS that are like the macros (for those who use a word procesor), is a program within the Cake/Sonar, this are user writable to suit your needs....you can Humanize an over quantized track, adjust volumes in EACH percution instrument within teh Drum Kits, control dynamics (velocity (VELPCT) porcentage that allows you to let say, lower the velocities without altering the dinamics, another is to make a track less monotonous, as an example, you have a track with the bass, and all the song has 127 note velocity, so, a bass player never plays at the same volume everynote, so, I use the VARYVEL (velocity variations), in wich you choose the range, lets say from 110 to 127, and in a split second, that track will randomly will make this variations, and so on. I thing is a wonderful software, a great tool.