Tips for MIDI users
(update 18/01/2000)

The MIDI language has been part of the studio worl for more than a decade now (they kicked in kinda slow, it took about 5 years for the benefits of the technology to strike the minds of studio owners), for automating effect program changes or sequencing instruments, even in some cases (for home-studio mixers) automating the mixer itself.

But this language is now too slow for it's original purpose. They didn't think that one day, the 10 fingers of the keyboard player would be replaced by hundreds of virtual fingers moved by a computer, now that these human finger movements have been changed to binary codes. In that case, it's true : MIDI is slow !

That's why a few rules need to be observed in order to respect the timing of the MIDI information. Most of these rules have been repeated over and over and over and over... I'm not saying I invented anything here.

Choose your tempo wisely

As soon as a sequencer is running, it emits a byte named 'MIDI Clock' 24 times per quarter note. This byte is prioritary over any other form of information, including the notes of your music, emitted on the same output.

Either you inhibit the output of MIDI Clock from within your sequencer (most of the time, you won't be needing it, except if you have outboard delay effects that are counting on that information to respect delay values, while your tempo is changing during the song. Nothing that a few strategically placed 'Control-Changes' can't solve !)

Either you keep emitting and in that case, you'd better work at a reasonable tempo value : between 50 and 180 BPM. Below that, there are too few timing informations for other machines (sequencers, arpeggiators, rhythm boxes, etc...) on the MIDI chain to correctly play semi or triple quavers (16th notes or more). Above that, you 'clog' the MIDI network with timing information, and it's the timing of your music, the notes trying to make their way through this mess, that's going to suffer. You can always divide or multiply the tempo by two, and apply the opposite to your sequences, can't you ?

Eliminate junk at the source !!!

Eliminate any information you don't need to be sent from your master keyboard, like aftertouch, at the source. If you play with that function enabled when you don't really need it, you're going to 'clog' the MIDI link between the keyboard and the sequencer, which will throw off the timing of your notes into the later. If you quantize like mad anyway, once in the sequencer, forget what I've just said.

Transfer to a tape machine

If you've decided to transfer the sounds of your sequences to a tape recorder (or Direct to Disk) rather that having them play live via the sequencer during the mix, it would be nice if the timing of the notes was perfect, wouldn't it ? Remember that in the MIDI world, chords do not exist. As the notes are sent one by one in the network, there are only rapid arpeggios. You have to count with about 1 ms for the transmission of one note (that's without the running status system, which can reduce an overall message quite a bit, about one third actually).

So, if your second verse starts with, simultaneously, a synth chord (5 fingers), a big piano chord (10 fingers), a bass with a sliding note (1 or 2 fingers + Pitch Bend info), a drum accent (BD + SN + Crash = 3 fingers), a guitar chord (5 fingers), and a few additional things......

That adds up to about 24 fingers, plus a whole bunch of other informations like maybe some aftertouch (see above) or other more useful stuff, even inevitable, like the bass' Pitch Bend. All of these things playing together means that there is about 30 ms between the first and last note of what is supposed to be everyone playing together, right ?

Unless you have severall individual MIDI outputs, on separate computer busses, like one can currently find nowadays with Steinberg, E-Magic, Opcode, MOTU, etc... (a MIDI splitter doesn't count, because it just repeats the same information on all the outputs ), you're better off passing the instruments one by one to the tape machine.

All you have to do is sync the sequencer to the tape machine once and for all, then to run the whole system several times, each time playing a specific track of the sequencer. You just have to mute all the other tracks of the sequencer (or use the MIDI Solo function, if your sequencer has one).

That way, the timing of the notes won't be thrown off by other parts which could be quite busy. Of course, it's up to you : if there are several slow chords, these can very well play along with other stuff. You should really concentrate more on the percussive elements of the sequence, which will greatly benefit from this MIDI network epuration.

Start with a countdown

It's really useful. It will enable the other musicians to get into the tempo confortably before playing the first notes (it's better to have two bars of countdown than one). It also enables your synthesizers to receive in advance all the Program Changes, Control Changes and System Exclusive messages that are necessary for them to start the sequence properly.

Everybody sing together now...

Avoid more than 3 machines chained in the THRU-->IN manner. Above that, there's a lot of chances you're going to end up with stuck notes, or notes that don't play. If you can't avoid chaining like that because you don't have a splitter, then place the machines that play the percussive sounds first, and the ones that play the lush, slow attack sounds last (I know, it's impossible to make such a choice now that all the machines are multi-timbral).