When I was very young (probably five or six years old) I wrote my first piece of music.

I spent several hours thoughtfully sketching out the notes of my composition onto a staff. I gave deep consideration to the pattens of notes in my piece, as well as to the recurrence of patterns and themes within its structure. I wanted to write something with clarity and depth. A thing of beauty.

I poured my heart into this composition, and my final rendition consisted of a full page of a music that I was very proud of. Though I was untrained as a musician at this point, I assumed that my hard work and creativity would be apparent when the piece was played by a competent performer.

A few days later, I went with my older brother to his piano lesson, where I asked his teacher to play my composition. To her immense credit, she gave it her best shot, faithfully playing every note that I had put on the page. And my heart sunk to realize that my composition had no musical value whatsoever. Where I thought I had been writing music, I had really only been drawing dots onto a page.

A few years later, when I was eight years old, I started taking piano lessons myself. And, like nearly all beginning musicians, my lessons were primarily technical. I learned to read musical notation on a staff. I played scales. And I played lots of simple little songs from a book with a red cover. As I progressed through this book, the songs introduced additional technical complications in both the reading of the music from the page and the movement of my fingers across the keys.

I took those lessons for three years, and most of the time I was very bored.

Like many other budding musicians, my practice sessions were spotty. I’d sit at the piano and stumble my way through a piece. The process was tedious, and it felt like a distraction from what I should really be focused on. In these practice sessions, playing the piano was all about rote memorization and repetition, and the whole notion of music was nowhere to be found.

Many of my practice sessions were squandered in unproductive improvisation. I noticed that, if I played only on the black keys, I could invent songs as I want along without ever producing the kind of discord that would result from the same kind of banging on the white keys (I learned later that I had actually discovered a pentatonic scale). As much as I enjoyed banging around on the piano, discovering music for myself, I don’t think my parents were amused.

And so they pushed me to practice my assignments. Given enough practice, I could usually convince my fingers to play the pieces in the book. But, since I never really felt like I was making music, I rarely had enough motivation to practice, and my weekly lessons devolved into a ritual of apologizing for my laziness and promising that I would be more faithful in my rehearsals.

When I was eleven, I quit.

But I didn’t lose my interest in making music.

When I was sixteen, I signed up for piano lessons again. This time, though, I chose an instructor who would focus on teaching music theory, ear training, and composition. I learned all about musical intervals and the construction of chords, scales, and modes. I learned to recognize the differences between a perfect fourth and a perfect fifth, and I got the point where I could identify any two-note interval just by listening.

As these lessons progressed, I wondered how we would approach the composition of any actual music. Since I was more interested in writing music than in playing it, and since I was interested in writing music for multiple instruments, I wondered how I would ever write anything more complex than I could play.

I never had the chance to find out. After a few months, my teacher moved to a new house a few hours away from where I lived, and he was disappointed to find out that I wasn’t willing to travel to his new place every week for my lesson.

Unable to find a replacement instructor to focus on theory and composition, I started taking lessons from a jazz pianist. I liked the fact that he taught me lots of theory, but I was frustrated that there was so much focus on performance (and improvisational performance), rather than on writing music.

In these lessons, the focus was again on performance, and on practicing the technical elements necessary for performance. The pattern from years earlier resurfaced, and my motivation to practice deteriorated.

Luckily, during the course of these lessons, I got the chance to play several interesting classical pieces, including Bach’s 2nd Invention in D Minor. I was fascinated by the layered counterpoint of this piece, and playing it reinforced my interest in the composition and construction of music.

A year or so later, I saved all my money from a summer job to buy an expensive polyphonic multi-timbral synthesizer with a built-in MIDI sequencer. Over the next few years, I wrote a handful of simple songs using that keyboard, but I was still frustrated with its limitations. The sequencer facilitated the recording and playback of live performances on the keyboard, but since I didn’t think much of myself as a performer, those capabilities weren’t of primary interest to me. The sequencer could also be manually programmed using a small LCD interface and a half-dozen or so buttons. With this rudimentary interface (which, by the way, provided no way to visualize the actual notes that had been saved in the current sequence), I programmed the entire Ouverture Miniature of the Nutcracker Suite, consisting of 18 instruments over 176 measures. My purpose in this exercise was to allow me to examine the orchestration in detail as I programmed it into the sequencer, to play back the piece focusing in on one instrument at a time, and to prepare me for the task of inputting my own compositions into the sequencer interface.

It was an interesting project, and I learned a lot about the music of the Nutcracker in the process, it still wasn’t a very good process for creating new music.

Over the last ten years, I have kept hoping that the computer software world would help facilitate my interest in writing music. And there now exists a plethora of musical software, but in my experience all of that software fits within one of these categories:

• Synthesizers — A synth is really a virtual instrument. It produces sound based on MIDI input. The sound source is usually some type of algorithmic oscillator with a number of adjustable parameters (waveform shape, envelope, etc). The texture and characteristic of the output sound depends on how those parameters are adjusted.
• Samplers — A sampler is really just a type of synthesizer in which the sound source comes from pre-recorded audio rather than from algorithmic oscillators. A sampler maps the pre-recorded audio data to the appropriate frequency and applies its own set of transformations on those sounds (such as envelopes).
• Wave Editors — This class of software allows the user to tinker with the actual waveform data of any sound, potentially by drawing waveform shapes on the screen or by filtering the data through various mathematical algorithms. Wave editors are often used as the source of sampler data, or during the design of synthesizers and effects processors.
• Effect Processors — An effect processor takes some source of sound as its input parameter and manipulates that sound using certain mathematical functions. Noise reduction, distortion, and graphic equalization all fit within this category.
• Sequencers — Essentially, a sequencer is just a tool for mechanizing the playback of music that has already been composed. Although these software packages come with an abundance of features making them suitable to the recreation of nearly every genre of music, they still don’t address the core creative process that takes place during the composition of a piece of music.
• Generatve Music Applications — These applications take some stream of random data and perform transformations on that data to create a sequence of patters. Then these patterns are piped through a series of rules to create a the notes of a composition. With this class of software, the user generally supplies a set of parameters to the rules, controlling the machine composition process. But the majority of the compositional task is handled algorithmically by the software, and the user really doesn’t do much in the way of musical composition.
• Multi-Track Recorders — This class of software records sound, either from acoustic instruments or from synthesizers, adjusting the balance of sound from each input source, and assigning those sounds to the appropriate left/right pan channels (among other things).

What I’m looking for is something substantially different. I’m looking for a piece of software that actually facilitates the composition of music. Something like this…

A Composer’s Workbench

Let’s say I’m at the very first stages of working on a new musical idea.

To get started, I create a new composition project, and within that project I start by jotting down a few musical themes. Maybe there are a couple of melodies that have been bouncing around in my mind, and for each of those melodies I create a new fragment of notation.

One of my fragments is five measures long, and I have a pretty good idea of the note progression and the rhythm, so I put those notes on a musical staff exactly as I’d expect them to be played. Another fragment is only three notes, and I don’t know exactly the type of rhythm I’d like to use, so I’ll just save this note progression without committing it onto a staff. I also have a great idea for a particular syncopated rhythm that I’d like to use somewhere in my composition, but I’m not sure what notes I’ll set to this rhythm, so I just save a rhythm fragment.

While putting together these fragments, the key of e-flat minor seems most appropriate, so I’ll assign that key as a property of my composition. Also, a time signature of 5/4 supports my underlying rhythm pretty well, so I’ll use that.

Now I’d like to start experimenting with my themes, so I start creating transformations of my existing fragments. For my primary melodic fragment, I create a melodic inversion. Anytime the original melody moves up by a major third, the inverted melody moves down by a major third. Anytime the original moves down by a perfect fifth, the inverted version moves up by a major fifth. There are several different methods for constructing melodic inversions. Maybe I’d like the constrain the inversion to the current key signature, or maybe I’m willing to allow accidentals wherever necessary.

As I experiment with these inversions, the software tells me that the inverted versions have a 86% conformance with the b-flat major key, so I assign b-flat major as a secondary key signature for my composition, transitioning into the primary key signature during an interlude section of the piece.

Also during this experimentation, the software suggests an overall chord progression of Ebm-Bbm-G-Cm as the underlying harmony for my primary melodic fragment, so I adopt that as a new harmonic fragment.

In addition to inversions, I create new fragments by transposing, reversing, repeating, slicing, stretching, and compressing my original themes. Along the way, I think of a few new themes unrelated to the originals, and I create fragments for those as well. The software allows me to tinker with the rhythms of my fragments and to combine the notes from one fragment with the rhythm of another fragment.

At this point, with a vast library of thematic fragments, I start working on some basic instrumentation, so I add a score sketch to my project. Perhaps I’d like to write a string quartet, so I add four instruments to this score: a cello, a viola, and two violins. The software knows the ranges, as well as the typical attack and decay speeds for each of these instruments (for example a cellist can’t play sixteenth notes at 120 bpm).

Now I begin laying down notes into the score. The notes come from my fragment library, and each fragment can be pasted as its own object into the score. Even after laying down the notes, I can still move fragments forward or backward by a few measures (or fractions of measures).

As I go through this task, I realize that my primary/secondary key signatures of Ebm/Bbm are slightly too high for the violins. They won’t be able to play the highest notes in my fragments. To solve the problem, I transpose the entire project down by three semitones, and my primary/secondary key signatures are now Cm/Gm. All of my other project fragments are also transposed accordingly.

After laying down a melodic fragment into one of the violin parts and a counterpoint fragment into the viola part, I overlay a harmonic fragment onto the composition and then ask the software to suggest notes for the cello that would work well with the melodic structure. The software provides several choices, and I listen to each of them, accepting many of the notes supplied by the software but rejecting a few that I don’t think work very well. This process gives me an idea for another harmonic progression that I’d like to use later in the composition, so I create a new harmonic fragment for it and lay that fragment down elsewhere throughout the composition.

At this point, I’ve pasted many of my fragments into the score, but the entire score is still pretty sparse. My main thematic fragments are in place throughout the first few dozen measures, where I want the first violin to play a half-speed rendition of the main theme, followed by a full-speed rendition of the secondary theme repeated twice. But then the melody moves into the cello part, followed by an aggressive inversion of the primary theme by the viola.

With this much written, I don’t yet have much supporting instrumentation for the non-melodic instruments, so I ask the software to start attempting to fill in those empty parts. In the background, the software creates thousands of combinations by inserting my thematic fragments (and their variations) into dozens of different positions. For each of these combinations, the software calculates the resultant harmonics and rejects the combinations that don’t fit within the harmonic parameters that I’ve defined. When it finds combinations that seem to work well in the structure of the composition, it suggests them to me. For each of those suggestions, I can reject or accept the new fragment placements into my composition.

In one section of the composition, the first and second violin each play the primary theme. But the second violin is a little delayed, playing the same notes one measure after they’re played by the first violin. At the same time, the viola plays the same theme at half-speed and transposed down by an octave. These three parts together create a very interesting counterpoint, as well as some harmonies that I hadn’t expected. I can select these measures and ask the software to extract the harmonic progression. That harmony can now be created as its own fragment and pasted elsewhere throughout the composition.

As the composition builds, the software keeps me notified of certain vertical and horizontal metrics. During an early legato section of the music, I’m averaging 2.7 notes per instrument per measure, gradually increasing to 3.1 notes per instrument per measure by the end of the 26th measure, but then the average abruptly increases to 7.2. Listening to the music at this point, I confirm that the transition really is a bit too abrupt, so rework a few of those measures to make the transition a little more gradual.

Anywhere where I’ve pasted a harmonic fragment, the software can tell me how closely the notes of my instrumentation match the acceptable notes for that harmony. In a few places, the actual instrumentation is slightly out of agreement with the harmonic fragments that I’ve created, but I like the texture, so I’ll leave it as-is.

The software also tells me that the most common melodic interval in my composition is a minor third and the most common harmonic interval is a perfect fifth. Of the notes in my composition, 94% are in the key of c-minor, 5% are in g-minor but not c-minor, and 1% are accidental to both of these keys. I’d like to add a bit more discord to my composition, so increase the number of accidentals, modulating some of my major sixth intervals into diminished sixths

This iterative process continues throughout the composition, until I’m pleased with the final result.

In other words: the software doesn’t compose music for me. But it provides a musical workbench for experimenting with ideas, using the common idioms of music theory to help me flesh out my original sketches and themes into rich, textured compositions.

Pros:

• There’s no software like this on the market. At least, I’ve never been able to find anything like it. Musical software usually only provides tools for synthesis or recording, but rarely provides tools for navigating the creative process.

Cons:

• This is one of my more esoteric ideas, and I actually hesitated before posting it here. it seems much more like a hobby project than a product idea.
• Development would be very difficult. It would almost certainly take me more than a year to develop anything remotely useful (even if I omitted most of the features that I’ve described above).
• Although I have some background in music, it’s been several years since I played an instrument at all, and it’s been over a decade since I’ve written any original music. So I’m pretty rusty, musically.
• I’ve never done any music/sound/audio programming, so I’d be starting out completely fresh in that field.

…

This is the seventeenth of 30 business ideas that I’ll be writing about over the course of 30 days. Or maybe it’ll only end up being 19 ideas in 36 days. We’ll see. Either way, one of these ideas will become a product over the next six months, and the foundation of my new software business.

This entry was posted on Saturday, July 15th, 2006 at 10:54 pm and is filed under biz. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.