Most of the music I enjoy is normally performed on actual instruments, and in most cases, acoustic instruments. Therefore I have avoided that whole realm of MIDI that uses completely synthetic sounds. So I spend lots of time trying to get the MIDI synth to sound reasonably like an actual instrument.

In some cases, it's just not possible to get General MIDI to even closely resemble a real instrument. For example, the single alto sax patch doesn't cut it; sax players normally use a whole range of tone colors in their playing, and we are used to saxes sounding like that. In other cases, you can get by, but there are things missing. A fiddle player will throw in some pizzicato or harmonics, but there aren't such patches available in General MIDI.

What I've found helps a lot is to study the actual instrument, and tweak the MIDI track accordingly. I did this first with the guitar, since I actually play, and know what sorts of things are done. So for things like slides, hammer-ons and pull-offs, chording, etc. I try to get acceptible General MIDI facsimiles. Using actual pitch range (non-players note: guitar music is played an octave lower than it is written!), real-world sustain, possible chords, and so on, makes the track more familiar to how we are used to guitars sounding, and hence more convincing. I've extended this to other instruments, which I don't actually play, by studying how they are tuned, studying the notated music, and listening closely. I know that there are REAL players out there who will find plenty of examples in my MIDI sequences where the part is impossible to really play. But the point is, the more devices you throw in that ARE actually used in playing, the more convincing it will be.

If you look at any of these files in MIDI sequencer software, you will notice that lots of "instruments" take up more than one track. Multi-tracking lets you get unisons in your chords, fiddle double-stops, or whatever. Or you can use pitch bends to just get that one guitar string, without changing the other strings in the chord. In the case of the South American charango, the mandolin-like instrument has 10 strings in 5 double-courses. One tuning for the charango goes: G5G5 C6C6 E5E6 A5A5 E6E6, so I need at least 3 tracks for all the unisons.

The virtual pedal steel needs 3 tracks. While a bar slide changes the pitch on all strings, the foot pedals and knee levers only change SOME of the strings. With the bar in constant position, you can get a variety of chord modulations just with levers and pedals. So I have labeled the 3 tracks 0,1,2 for pitch changes of no change, one semitone, and two semitones. In addition, the pitch bend range is widened from the default ± 2 semitones to ± 8 semitones, or even greater (like for those loooong slides in "Sleepwalk").

The virtual koto uses 2 tracks for notating the traditional ornaments. One such is playing a unison by bending a lower string up to the pitch of the next unbent string. The "shushing" sound is an attempt to convey chirashizume, where the plectrum is scraped along a string lengthwise, an ornament commonly found at the end of a piece.