Sound files will be added as time (and free web hosting space) permits.
In order to demonstrate the organ on something other than simple hymns, I used the Artisan Instruments controller's MIDI input capability to play the organ.
Many modern or rebuilt organs have this capability; but I didn't want to just capture a performance (since I am not an organist and I had no one available who could play the demanding literature I wanted to use), but instead to create my own interpretation from the original music. I also wanted to keep faithful to the registration suggested for each piece, and to play each staff on the division(s) marked, changing it whenever marked on the score.
All recordings on this page were made digitally on my laptop computer (P-4 3.3Ghz under Win XP) using a USB digital audio interface capturing 16 bit stereo at 44.1Khz sample rate (CD quality). The files were then compressed to MP3 using Nero 7's recoding function. The microphones (old but superb Radio Shack PZM condenser) were placed on either side of the center of the empty church. The organ was played by the same computer using a USB MIDI interface. In fact, the laptop was running Overture playing the score, with my custom MIDI filter application interpreting the registration, division and expression changes, and the digital audio recording program, all simultaneously with no problem.
No artificial audio processing such as filters or fake reverb has been applied to the recordings. Although the MP3's are not as good as the original recordings, especially on the higher frequencies, they are remarkable considering the file size is only about one-twentieth of the original!
Unfortunately, I only had a limited time to complete the recordings, and as luck would have it that afternoon was picked by workers to move stuff around inside the church, so in a few places -- in the Vierne Sixth Symphony for example -- you can in places hear keys jangling, doors slamming, etc. Also, in order to easily make the recording by myself I used automatic gain control which had the unfortunate effect of reducing the volume change effect of the shades. I hope to polish up some things and make new recordings but I wanted to have some samples up now.
Symphony No. 4 for Organ
Symphony No. 6 for Organ
Symphony No. 4 for Organ
Trois Préludes et Fugues
The music I had was either from printed scores I inherited from my mother (including some of my favorite Messiaen and Langlais pieces), or from a CD I bought that had hundreds of public-domain scores, including all the works of JSB, the complete Vierne and Widor symphonies, and hundreds of other works.
Using an Optical Music Recognition program called SharpEye 2, I converted the graphic images of the music to MusicXML files that I could import into my notation program, Overture 4. Even though I had to correct the inevitable recognition errors, this still saved me many dozens of hours compared with manually entering the score.
The next step was to go through the score and adjust the note lengths, first making all notes legato, then shortening ones that were repeated, eliminating tied repeated notes, etc., according to standard organ articulation rules. This was the most tedious step, but was absolutely necessary for decent sound. There are still a few spots in the samples I need to fix, since it sounds different on the real organ from listening to synthesized output on the computer. I must say both the MIDI processing and the Moller electropneumatic action give very quick response, as can be heard in the Vierne scherzo and the Dupre prelude.
Tempo changes were next, listening to the output until I was satisfied.
Incidentally, Overture has a feature whereby note start/stop times, length and tempo values can be slightly modified randomly to create a more human (read: imprecise) playing, but I have not had time to experiment with this yet.
After the piece played OK on the computer, I inserted the registration change points, expression shoe changes and division assignments (more on this below) and was ready to work out the registration on the pipe organ itself.
After much experimentation, I had settled on a notation program called Overture, since it had the best graphical MIDI editing of any of the advanced true notation packages I used. There are many fine shareware and freeware MIDI editors as well as commercial programs that handle MIDI data, but all of them (even Overture) are lacking in certain essential aspects.
For example, in all of the notation programs I tried, each staff (or voice within a staff) is assigned a MIDI channel and that corresponds to a division on the organ. So even though packages like Overture allow eight voices per staff, it is awkward to assign channels, since in order to notate complex measures you may need to use two or three voices simultaneously on each staff. Moreover, organ music may change divisions on a staff frequently, in some cases several times per measure. This means that correcting mistakes means re-channeling the individual notes manually. Of course, as the piece plays, these divisional changes are not easily understood while watching the music on the screen.
Another problem is that notation packages are geared towards entering notes, which are transcribed as MIDI Key On and Key Off messages. But you cannot, in any of the programs I could find, enter Artisan registration data (which is just a Key On message without matching Key Off) except by making a separate registration track with dozens of ugly, insanely complex chords tied together over long stretches of measures. Changing registration this way is also impossible for a person to follow or easily change. I decided there MUST be a better way!
Using an excellent no-cost open-source MIDI framework called MidiShare, I wrote a filter program in C++ that accepts the MIDI output from Overture, passing the note data but intercepting controllers. I defined unused controller messages for registration, re-channeling and expression shoe data. I then made a user interface that allows the easy setup of registration as if the registration controller value were a particular combination action setting.
Within Overture, simply click a registration combination on the palette (I used Overture's customizable libraries to predefine the necessary controller numbers) and click it where you want it on the music. You can attach it exactly at a particular note to make registration changes within a measure.
Once I have the notation done, I play parts of the piece at each registration change point and experiment to find the exact sound I want. Then. in the filter program, I set up the stops and coupler checkboxes for that number combination. Once I have gone through the whole piece and finalized the settings, I save the registration file for later use.
Similarly, the two expression shoe levels and -- in the unlikely event it is necessary -- the crescendo shoe level, are changed by using the appropriate MIDI controller with the value parameter being the shoe level. Smooth volume changes are easily entered in seconds using Overture's controller graphic window and its tools.
Finally, the problem of re-channeling was solved by first assigning an unused MIDI channel to each staff in Overture, then using another of Overture's customizable Expressions Palette with text values of the standard organ notation such as Ped. R, P, GPR etc. and clicking first the appropriate palette expression and then the note on the staff where you want the division changed. The filter program automatically reads the controller number for that expression as the music is played, and cancels any pending notes playing on the current division at that point and restarts them on the new division, and re-channels all future notes from that staff to the new division until the next change is received. It also sends the appropriate intermanual coupler on or off commands to the organ depending on the controller number.
All this sounds complicated at first , but once you understand the system it is really simple and works very well. It is repeatable, and does not require human intervention while the organ is playing as did my original hand registration method (which was used for some of the first pieces recorded).
This page was last updated on 11/10/08.
All content copyright © 2006 Brian F. Bailey, W4OLF