Select your language



Composing with virtual instruments

In 2015 I started to study music composition seriously. I am currently working on adaptations of melodies from the renaissance, baroque and classical periods. Longer term I hope to progress to the creation of compositions in the sonata form.

My work is only relevant to work produced primarily from 1500 to 1900, using most classical conventions.

What do I do?

At the first stage I am learning the skills that a composer learned at the beginning of the 18th Century.

For this I choose familiar musical selections. I analyse their structure and I experiment with various variations to understand the essence of the melody. This inspires new melodies in the original style that I can cast into a familiar style for a subsequent performance of that or of a similar piece of music.

But the great difference arrives in the following steps.

All my work is done on "virtual instruments", which is not always easy.

At the present time, only some instruments are readily available for realistic emulation. These are the common keyboards and certain woodwind, percussion and plucked instruments. In renaissance to classical use, frequent instruments have been harpsichords, pianos, harps, guitars/lutes, flutes, and some wind instruments. Some adaptations to organs are also available. In common use, artificial violins, violas, violoncellos, horns, trumpets and human voices remain less than satisfactory, without special local modifications.

The second part is "mastering". This is a delicate and relatively technical procedure. I took many years to understand and to apply every aspect correctly.

As a first step, written scores produced by MuseScore, Finale, Sibelius or Dorico often need further work in velocity and rhythm. Specifications of dynamics (mp, mf, f etc.) and tempo (often in quarter or half notes) must be provided for every part of the score. Also while working with written scores, I use transpositions that are more suitable to my instruments or the original composition.

After that my work proceeds to a digital audio workstation that transforms the music for public consumption (a DAW used for "mastering a recording"). Many wish to automatize this step, but so far the results have been disappointing (e.g. in MuseScore 4).

A DAW provides these principal enhancements: (1) geographical placements of the instruments within your spatial acoustic placement, (2) enhancements to frequency domains that promote some notable and preferred regions, while keeping other regions sufficiently audible, (3) adequately adjusting the duration of the time of notes, in conjunction with local dynamics, (4) lenthening or reducing note durations between notes, so as to provide a credible tonal landscape, (5) appropriate reverb specifications between "dry" and "wet". "Wet audio" refers to a processed signal used to provide desired depth and distance to your instrument, while "dry audio" is the original untreated signal, (6) and overtone protection, in order to reduce the chances of extremely high or extremely low tones.

So far I have found no instrument that can perform all these manipulations automatically, and so I perform these adjustments separately for each piece of a composition. That in short describes the work I do for each composition.

Why? is a totally free service that I promote for the following reasons.

1. Novel musical creations. For example, the original melodies of Turlough O'Carolan were short tunes, often less than one page. From that I created and published a set of complete compositions.

2. Lesser known meritorious compositions. Examples on my web site have been: the earliest compositions by Antonio Vivaldi, highly selected sonatas from among 555 works by Domenico Scarlatti, and rarely heard works by John Dowland. Currently work is under way to produce a series of works by Fernando Sor.

3. Use of the best available web-capable virtual instruments. For written musical scores are read, analysed and then implemented in musical form. This often requires much work and experimentation before a pleasant performance is possible. uses the best available musical equipment.

4. Mastering. Since works from written scores, a detailed preparation is required. This is quite different from traditional music recording. In our case, the instruments must translate a series of musical symbols into audible text. This involves a series of serial dependencies that is different for the production of one, two or three voices, as against the realization of an orchestral group.

Our instruments are rather limited. Sound recordings of the lute in broadcasting, for example, can use special microphone sets, while there are no usable instruments for simulating a virtual lute. We must use instruments simulating a classical guitar, and after extensive searches, I am limited to just one type of guitar that can be used for our simulations.

5. Future. Our style is then limited to a certain set of tools that are compatible with today's technology. This is what we are currently using in Will this change radically in the near future? No, even if we had access much more resources, this is unlikely to change a great deal. Given the current situation, major changes in virtual technology are more probable in about ten years from now.

Let's enjoy what we have now.


Eric Keller

Responsabilities 1978-2008: