Alright, so I chose to do the patches with the DX7 Librarian, because FM8 uses a completely different terminology (for example using real time units instead of the DX7's 0-99 steps). Unfortunately the DX7 Librarian is a bit fiddly to use: one cannot just type numbers but everything has to be done with the mouse and the sliders are rather small. 

What's worse, some numbers it would simply not take and jump for example from 38 to 40. 39 was unreachable, and so were quite a few numbers.

However, the main issue is somewhere else: each DX7 patch has about 120 parameters (not counting the global "Function" parameters, but only the actually patch settings). There are 36 patches for both synths in Vision de la cité interdite. That's over 4000 parameters that I had to dial in on these tiny little sliders. I have done boring stuff before, but this was bad. 

It took a long time to do this. Interestingly, now that I have a DX7 standing next to me, it would have been much easier to use the often criticized interface of the synth to do it: right hand chooses the parameter, left hand moves the data entry slider, done. I would think it's five times faster doing it on the hardware synth. I do not think it would have been faster using Dexed of FM8 because they all need the mouse, and that just takes much more time. Note to self: maybe I should have attached a MIDI controller, mapped the 120 controls to it and went from there. More time needed at the start but probably well worth it.

Knowing that I didn't have all the numbers right, and knowing that Sascha Lino Lemke had used the same program for the patches for Atlantys the next horror already approached my door: printing out the DX7 Librarian patches, comparing them with the patches printed in the score and correcting them using Dexed. Argh.

Here's how the patches look like in the score (it's quite a nasty print and sometimes I were not really sure if it's a 3 or an 8 we're talking):

 

And this is one of the 50 patches that had to be corrected:

 

 

What we are talking here is checking over 6000 parameters by hand for both Atlantys and Vision de la cité interdite. Why do I have to do this?? Should I not rather practice the music? Must be my karma or something. Think of me as a life wasted.

Anyhow, here's the goodies: all the patches for both pieces, one bank for each synth. I take no guarantee that everything is correct, but unless somebody checks all of this again this is probably as good as it gets. The zip-file contains the two SysEx-files and a read me with more explanations.

Murail_Atlantis_Vision_DX7SysEx.zip

Only now, nearly three months later had I finished the first item on my ToDo-list. Finally I could start play the music and actually listen to the patches and compare Dexed and FM8. They sounded slightly different, especially with some sounds using vibrato, but all in all they sounded alright and were a good basis to start working. Off to Brussels for rehearsals with Frederik Croene.

Entrance Jean-Luc Pluvier, who was most helpful by providing scans of the patches and giving me hints on some peculiarities of the DX7 and how they were employed in the pieces.

What one needs to know when playing the two pieces:

The main thing is: the DX7 makes strange sounds when changing from one program to the next. As opposed to all the emulations I came across the DX7 does not stop the sound when changing a program but keeps on playing the envelopes, albeit with different algorithms. I have yet to find out what exactly is happening, however, it is something that Tristan Murail makes use of. In the score, program changes are depicted as numbers in squares. However, some of these squares also feature an arrow which is meant to say that the program change has to happen at exactly this time. This may or may not result in the strange sound and it is also not quite clear to me, where exactly these changes should occur. It should say it in the score, but I really don't want to say it's a bad score, indeed it is a very good one: everything is there, the patches, annotations on certain passages of the piece, annotations on the notation, on the piece itself, on the global parameters of the synths and how they change when playing both pieces. It is a good score! Unfortunately this little detail is missing and unfortunately it's an important detail. So, two more points on my ToDo list. 

- What is happening on the DX7 when programs are changing?

- Where exactly is it happening?

Another thing which confused me was the usage of the modulation wheel. Normally a modulation wheel would have no effect in the down position (no vibrato) and full effect in the up position (Brangäne, Wagner: Tristan und Isolde). This is true on the DX7 as well as far as pitch and amplitude settings go, however, when used for the EG bias, which is the case in Tristan Murail's pieces, its the other way round. (The EG bias controls the amount of modulation the modulator has on the carrier. When the modulation wheel is down, the modulation amount is also down, meaning that one will hear the pure sine wave of the carrier. When the modulation wheel is up the modulation amount is up (up until the programmed amount, that is), which has the effect that the sound changes from the sine wave to an overtone rich sound. So, musically this makes complete sense: changing the modulation wheel makes the sound richer, but it can be confusiong to read that an effect has to be on all the time.) If all of this is gobbledigook to you: Don't worry. Just keep the modulation wheel in Vision de la cité interdite in the up position all the time. If you are hearing a tremolo or vibrato effect though, something is wrong in your master settings. 

Basically, the important thing that's not written in the score are the sudden changes in the sound when changing from one program to the other. I have yet to find out exactly where they are. When I know, I'll post it in part IV of this little series.

Back to the timeline though. For the first rehearsals with Frederik we used a basic setup with Dexed, since Frederik doesn't have Native Instrument's FM8 and I wanted to enable him to practice with the sounds. In the rehearsal sessions quite a few more questions came up. Here's a copy of the excel file with what I had to do:

It was mid-December and I was not even close to having a working setup. I just knew that there was still a lot to do over Christmas.