How a complete piece for melody and harmony instrument comes about – from first ideas to finished score : Selah for violin and piano : before the first note – constraints and conditions : creating melodic phrases by vector mapping to pitch : working with a tonality – the double harmonic : using the attributes parameter in Opusmodus to include ornamentation : generating chords from melodic phrases : the rough-prototype : Selah for cello and piano : similarities and differences.
In Part 1 the focus was on how the different parameters of note-length, pitch, dynamics, and structure came together in code-specific programming to compose music. Although there were many examples taken from compositions created entirely through CAC the whole narrative of a piece’s composition was not yet up for discussion. It was felt that the making of musical phrases, sentences and paragraphs had to come before the story of a composition might be told.
In Part 2 the emphasis in many of the sections will be how a complete piece comes about: following first ideas to final score. The example scores are not wildly experimental, indeed they play safe, if only for the reason that they were mostly commissions and had to be playable! It is this factor, connecting with the real-world of music composition and performance, that computer-assisted composition presents its greatest challenge, and where pragmatism and common-sense may need to be applied simply to get the project completed on time! Unless composing time is not governed by completion and rehearsal dates, only part of a composition project may be feasible for a composer in the early stages of working with a CAC system.
No matter what the style, genre or system a composition owns, there are certain universals that belong to the composing act. No matter where you are musically Melody and Accompaniment is likely to figure somewhere in most composing situations at some time. Often this is a necessity driven by the nature of the scoring chosen (or required). In creating a piece for violin and piano there is an assumption that these instruments take particular roles of melody and accompaniment. But there are also many variants of this form. It doesn’t have to be melody instrument with an accompanying harmony instrument in a subservient role. Mozart wrote piano sonatas with optional violin obligati, which is turning the idea on its head. The word dialogue can be a powerful clue as to how a composition having melody and accompaniment is often be composed. In Roberto Gerhard’s remarkable Gemini there’s a fine example of how violin and piano come together as though they were a single instrument, eschewing the idea of melody and accompaniment altogether. Whatever direction taken at the outset, it’s probable that melody will share some aspects of its material with accompaniment, and the most usual of these shared parameters is pitch, in the form of a scale or pitch-row.
In 2012 I was commissioned to compose music for two teenage musicians and their parents, both professional musicians: to celebrate a Bar and Bat Mitzvah. I had already had a little experience of composing to a Jewish subject in a setting of the Seven Nuptial Blessings for piano duet. I had made no attempt in that work to make reference to Jewish music, in fact it was Stravinsky’s Les Noces (featuring the sounds of the Russian Orthodox rite) that eventually left its mark on these seven duets.
I had a Jewish Prayer book and made myself familiar with the Mitzvah ceremony, which takes place as part of the morning service for Shabat. Rereading the prayers I became aware of the word Selah, a kind of heightened Amen, a word that occurs over seventy times within the Psalms. I began to take in Jewish numerology associated with this festival of welcome into adulthood, in particular the Hebrew gematria, and the numerical equivalent of the Hebrew word for life, ‘Chai’. I knew the instruments I had to compose for and there were plenty of musical settings of the important psalms and hymns sung at Shabat freely available on the web. I found one in particular that intrigued me, a setting of the Adon Olam performed by a Jewish-Iraqi ensemble from Bagdad. I decided before a note had been written that I would use this well-known melody as the basis for at least one of the two Selah. The other would carry something of the spirit of this melody but be more meditative in aspect. The tonality of the version I found of the Adon Olam was that of the Double Harmonic scale.
Notice that I have extended the compass of the scale to enable me to bring in the all-important supertonic and mediant, both featured in the Adon Olam. I’ve cast the scale for violin, choosing a transposition that includes three open strings. At this point the numerology factor appears. I have two pieces to compose, and eighteen seems the magic number (Bar and Bat Mitzvah candidates traditionally receive gifts of money in multiples of this number). So the number nine was chosen to drive the sectioning of the pieces.
This doesn’t look promising because there seems just too many repeated pitches. But as we saw in the section ‘Starting with Pitch’ there’s an ornamentation function (and introduced here a vector-smooth function) able to take out some or all of the repeated pitches. This series of motifs looks a whole lot more promising! But, an errant low f# has crept into the output, outside the violin’s range. Using the ambitus function easily adjusts that ‘error’, or we could simply transpose those phrases with the lower f# up an octave. In fact this would give, in general, a more authentic feel to the phrase collection.
The idea of tonality-map is to convert a chromatic stream from the white-noise fractal collected in the variable motifs-x into the tonality of the double-harmonic with its root on g3 (lowest open string on the violin).
Finally, let’s take out some repeated pitches to create rhythmic shapes. Again, this is a process using those binary functions we’ve met before in the section ‘Starting with Pitch’:
It’s this kind of playful experimentation that a CAC system is so good at providing for the composer. If these options are collected as notation snippets there’s now wealth of possibility available to be scanned and reviewed, and from which to make the next move.
Listening to the recording of the Ensemble Bagdad-Jerusalem playing in a garden in Fes in 2010, their version of the Adon Olam begins with a short violin improvisation rich in ornamentation; nothing elaborate, and nothing that could not be described with the standard ornaments of Baroque practice. Ornaments are powerful devices to colour and enrich melody, but are often difficult to realise in computer-aided systems. As the starting point for Selah is to be melody, adding ornaments to the double harmonic tonality contributes powerfully to the ‘sense’ and direction of each motif.
The Opusmodus system is unique in having, in-built, an extensive notation for ornaments. These are notated within the linear script of OMN (OpusModusNotation) already described in early sections of this text. Up to this point this script has focused on Length, Pitch and Dynamics. To add ornamentation, indeed a whole raft of musical detail from Bartok pizzicato to tremolo and arpeggiation, Attributes can be added as the fourth parameter of the script. Here’s the first of the nine motifs now termed the variable Phrase-1:
Whilst a library of traditional ornaments can be invaluable to support a score like Selah which references Jewish music and in which ornaments are common-place, composers continue to use ornaments in modernist and art music scores as a means of melodic intensification. For solo violin go no further than Gyorgy Kurtag and his Kafka Fragments Op.24, but listen Stockhausen’s Mantra for two pianos for a mix of traditional and newly-composed ornamental textures.
The major issue with newly devised ornamentation is acknowledging exactly where in the ornament is to be placed: before or after a metrical target. In the section ‘More on Rhythm’ there are examples from my work String Trio (2012) of how a CAC can invent and realise ornaments with a speculative degree of random choice of both pitch and rhythm.
Moving on to the piano accompaniment for Selah, in the score-script each phrase is tackled separately. The first action is to strip out both the Attributes of ornamentation and note-length and focus on pitch. This is can be done most elegantly in OMN.
As in earlier examples we’ll take this expression apart line by line starting at the end of the expression:
- (gen-repeat 2 (rnd-order pk-1 :seed 2))
- => (cs4 e4 g3 g4 bb4 d4 fs4 cs4 g4 d4 bb3 e4 cs4 e4 g3 g4 bb4 d4 fs4 cs4 g4 d4 bb3 e4) ; this is the material we need to generate chords
- (gen-chord 36 2 4 -18 12 (gen-repeat 2 (rnd-order pk-1 :seed 2)) :seed 781)
The gen-chord function has been set to range of three octaves (36), to create chords of between two and four notes, and transpose between minus 18 and plus 12. This is the result: twenty-four pitches arranged into eight chords.
Here’s the full expression headed by tonality-map. This will act upon the output shown above and manoeuvre the output into the double-harmonic tonality to be played by the right hand of the piano referring to this :ambitus ‘(g3 d5). Here below is the outcome:
Building the expression k-1 towards the draft of the right-hand part we can begin to place it as an accompaniment to the phrase-1 of melody. The melody is nine pitches long and in 3/4 metre, so now a rhythm for the accompaniment needs to be invented.
There’s an approximate piano accompaniment here, but when it’s put against the violin phrase it is not adequate. The voicing and octave placement is not satisfactory. The final version must be altered by hand on the score-writer. The CAC version is fairly close, but without substantial extra coding, does not stand up to what a composer will intuitively require in such circumstances. The CAC version must be considered a rough prototype. It gives the sense but not the detail of Selah. But it does allow the composer to use the code (carefully explained here) as a possible model for composing the remaining eight phrases, which indeed is pretty much what happens.
To emphasise this point, let’s see the entire code for Phrase-9, the final phrase. All the functions should be familiar and many of the values that fill the argument slots in each function are retained, for example those in tonality-map and gen-chord, are identical to the previous phrases.
The second Selah is for violoncello and piano and is not so focused on melody and harmonic accompaniment. It is a largely heterophonic piece played out mostly in 3 separate voices. There are, as in the first Selah, phrases created in the same way, but each phrase is treated more like a sequence of variations than as continuous and evolving melody with a harmonic underlay. It’s lively and rhythmic, and each phrase carries a link to the next. The piece ends with a rendition of the original Adon Olam melody as Phrase-9. Here below is just the cello part preceded by three bars of its link.
In Phrases 6 & 8 there are examples of the melody instrument taking the accompanying role. In Phrase 6 the cello plays a slow bass-line while the piano plays a melody in the left hand, chords in the right.
What hasn’t been shown to date is how a large collection of sections (Phrases in Selah) are collected together and ‘assembled’ with the function assemble-seq. This is a good illustration of how important it is to name sections coherently.
As with all the compositions in Part 2 the reader can download both full-score and score-scripts of Selah from the composer’s website archive.