My Midi Sequencing Techniques

All my midi sequences are produced by playing the notes on a Yamaha Clavinova keyboard. I find that for the most musical and realistic effect this works well, especially for the romantic repertoire where plenty of expression and rubato is required.

I have been developing midi sequences for about two years now. Most of them are for piano or organ, and during the last six months I have been experimenting with midi orchestration using a Roland JV1080 and Cakewalk 3.0 software. I have acquired several techniques for improving the quality of my sequences and for making them more musical.

I play the piece through to the end or work at it in sections which I paste together, and then use midi editing techniques to improve upon what I have recorded. I examine the midi events list using Midisoft Studio or Cakewalk and correct or eliminate wrong notes, insert missing notes, adjust the velocity values to improve the melodic line, phrasing and harmony, and adjust the note position and tempo to improve the rhythm, phrasing and expression.

I listen critically to the playback and try to make the music sound as I would like it to during a live recital. I add reverberation to make the piece sound as if it were being played in a concert auditorium.

One of the problems in playing an electric keyboard is that the action seems to be more uneven and has a different feel to it when compared with an acoustic piano. During playback of a sequence, a note may suddenly sound too loud or too soft. This is also partly due to problems with finger technique. To compensate for this I adjust the value of the note velocity to achieve a smooth legato.

In the midi orchestration of piano concerti, I use the keyboard to play the orchestral tuttis as well as the solo part. In a sequence where the piano is played real time, it is not possible to fit the orchestral part to it by step recording methods since the rhythm and tempo are not rigidly linked to each measure. I examine the midi event list for the piano part, and using an orchestral score, determine where each note of each line of the orchestra coincides with the piano part.

Sometimes I have to calculate the midi time locations of the notes and have to shift the parts slightly early or late in time to take into account the attack time of the synthesizer patches. When I have put in all the orchestral parts I balance each line for the correct audio volume so that they blend together well and that no line drowns out solo part or other instrumental parts. I add varying expression controllers to make crescendos and diminuendos. I also add pan values to create the spatial stereo effect of the different sections of the orchestra on the concert stage.

It is very challenging to make a sequence with orchestral accompaniment sound convincing. Some of the patches on even good synthesizers sound rather like an electric organ or a gazoo. Some solo string patches sound terrible while some string ensemble, woodwind and brass patches are acceptable. There are probably over a hundred ways to play a single note on a violin depending on bow position, bow pressure, velocity, vibrato etc. I don't believe that it is possible for a synthesizer to do this, but maybe it will be in the future.

This is extremely time consuming but very rewarding, and lots of fun. My most ambitious project so far had been the first movement of the Tchaikovsky Piano Concerto. I have also sequenced the Chopin 2nd concerto 2nd movement, the Rachmaninoff 2nd concerto 2nd movement, and the Scherzo from Litolff's Concerto Symphonique.

These orchestral sequences require a good quality synthesizer with 64 voice polyphony such as the JV1080 for best results. The Tchaikovsky concerto pushes the limit of polyphony, and occasionally a few notes drop out. It is likely that the playback may sound different on other synthesizers or sound cards due to the differences in audio levels of the patches. Adjustments can be made to compensate for these differences.

Return To Home Page