X. Orchestration

Transcription from piano

Mark Gotham

Some orchestral composers write for piano first, others write directly for orchestra,[1], while others still (perhaps most) sketch scores in a short-score format to begin with that is neither one nor the other.

While composers use a variety of approaches, it’ll be most useful for us to focus on the first task: orchestrating from piano music. Among the reasons for this:

  • there exists so much wonderful music in both piano and orchestral forms;
  • it makes you consider closely the similarities and differences between two contexts.

It’s best to think of this task in terms of going back to core (‘pre-pianistic’) idea and re-writing, re-realising that idea for orchestra. You might like to think of this as analogous to translation, rather than mere transcription.

This chapter will begin with some basic principles, and then move on to some magnificent examples of piano-orchestral pairs in the repertoire. I’ll finish by putting my own neck on the block, going through an orchestration of my own after a piano piece by Bartók.

Basic principles

Firstly, start by reading the original closely – really closely. Ask yourself:

  • What is the piano effect trying to create? How might this (or something similar) be best achieved by the orchestra?
  • What extra rhythms – and perhaps parts – could be extracted from the harmony, register, accents.

Examples for extracting ‘additional’ material from piano music include:

  • Extending notes: if the piano’s sustain pedal is down, you may wish to add parts sustaining each of the notes beyond their notated length (to the notated length of the pedal changes);
  • Extracting accents: accents of all kinds can be separated off into separate parts in the manner of the ‘attack-sustain’ effect discussed in the previous chapter: some parts play the music ‘as written’, others provide the attack only.

Examples that require re-working include:

  • Oscillating figures such as broken octaves. This is a pianistic solution that usually needs re-working for other instrumental forces. Many other instruments can repeat notes more readily than the piano. (Don’t forget to consider the register and rhythmic levels; as a rule of thumb, maintain at least those of the original.)
  • Arpeggiation. Large florid arppegiations by constrast are better suited to the piano than most other instruments. You will likely want to break them up into multiple parts, dovetailed together to give the continuous impression of the original.

Brief Examples

Let’s see this in practice with two examples from the piano repertoire. Consider how you would orchestrate them; then compare your result with the possible solutions below (but remember that there are always many different viable ways to do this – the ‘solutions’ are just examples). Plan on realizing the Chopin example for string quartet, and the Beethoven for classical orchestra.

Chopin: Prelude op.28 no.13

Beethoven op.13

In the Chopin prelude, the left-hand distribution invites re-working for orchestral instruments. Here is solution that re-distributes the material over two parts (Vla+VC). Even if you decide to do some kind of re-distribution in principle, there are a lot of different ways to realize it. I like this particular solution because it separates ‘main’ chord notes (cello) a broken thirds figure (viola) and in so doing, it brings out the very slight metrical dissonance hinted at in the original (here: 12/8 in the cello vs 6/4 in all other parts).

You may have noticed an elephant in the room here – I’ve changed the key. Whether and when it is appropriate to do this depends on context. For an orchestration of all 24 preludes it’d be best to leave the keys as they are; in doing just one or two, we’re freer to make changes likes this move from F# to F in order to ‘fit’ better on the string instruments, including hitting more open strings in the cello (which helps make some of the more unwieldly lines later on more practical).

Incidentally, earlier music is more likely to impel you to change keys between keyboard and orchestral versions. At least Baroque and Classical composers generally used a narrower set of keys for their orchestral works than they did in keyboard music.

Turning to the Beethoven, we have broken octaves in the left hand, a variable number of chords notes in the right, a crescendo, a growing tessitura, and some sfz chords. Here is a solution with viola, cello, bass, and timpani sharing that C pedal across three octaves and three different pulse values. Note also:

  • Where parts are added: first on the sfz, then twice in the next measure on the basis of motivic parallels and accelerating the rate of change;
  • The use of horns in Eb and trumpets C – a common device for composers of the time working with ‘natural’ brass instruments (see the 5th symphony, for instance).

Transcription case-studies

Now let’s turn to some longer-range examples, including some of the most celebrated repertoire examples, and some specific questions:

  1. Mahler 4 / iv (piano and orchestral versions both by Mahler).
  2. Bach: Partita BWV1006 (for violin) orchestrated by Bach as Cantata 29 movt 1
  3. Mussorgsky: ‘Gnomus’ from Pictures at an Exhibition (orch. Ravel and Stokowski)
  4. Stravinsky: ‘Danse Sacrale’ from the Rite of Spring (piano and orchestral versions both by Stravinsky)
  5. Bartók: VIII from Eight Improvisations on Hungarian Peasant Songs (orch. Gotham)

Mahler 4 / iv (piano and orchestral versions both by Mahler).

Question: How variously might you orchestrate ‘the same’ material.

Here are some extracts from Mahler 4 / iv with red boxes highlighting orchestrations of the arpeggiated accompaniment pattern. The extracts show Mahler creating a set of timbral variations on a single, simple idea. (Note that the pages are each separate extracts from the movement; they don’t run continuously together.)

As so often in Mahler, this music began life as a song forming part of a song cycle for voice and piano and was only later integrated into this symphony. The harp features in each of these extracts and is almost identical to the equivalent part of the original piano accompaniment, so we have that point of reference on the score and can see what else Mahler does with it.

  • At the start (first page, m.1) we have a rather lovely distribution that’s a long way from the simplest, easiest option: the violas take the D3 alone, while the cellos take the short, fast B4-G2 leap across a long glissando. Note further difference from the harp such as:
    • all the dynamic and articulation markings added in those string parts;
    • the cohesive elements: overlapping the viola and VC parts, and the horn pedal on D4 (sounding pitch).
  • In measure 4 (still on the first page): We start with clarinet 2 doubling the harp exactly, before moving to …
  • Rehearsal Mark 1, with a distribution of the harp part on to the Bass clarinet and clarinet 2. This distribution resembles the opening in use of similarly paired-parts (Vla and VC; B.Cl. and Cl.2), and in the separation of the rhythm. The pitch distribution has changed (now ordered G2-D3-B3), and the expressive glissando is gone. Cellos ‘double’ the G and D, but on the beats, thus slightly ‘missing’ the original D and creating a slightly heterophonic effect.
  • Later in the movement (on the next page, marked p.341 in this score and at 3’28” in the recording below) we see the cellos double the harp part directly. This G2-D3-B3 version allows cellos to use two open strings, which makes for a gentle, easy spread that befits the mood. The Cor Anglais now picks up the D4 pedal.
  • From Rehearsal Mark 12 (5’19”) the harp moves to a lower tessitura as part of the ‘winding down’ at the end of this symphony. Accordingly, we move from cellos down to double basses who double the downbeats much as the cellos did at figure 1, and again with an open string available (one this time), but rhythmically simpler in this case, without the heterophonic effect.
  • On the next and final page, you can see that that harp-bass doubling, and that same low E1 pitch (double bass open string) mark the final sonority of the symphony.

Bach: Partita BWV1006 (for violin) orchestrated as Cantata 29 movt 1. (orch. Bach)

This page provides scores for the violin part. Consider the following questions:

Question: How much can you get out of one single line? What is appropriate to draw out of the original, or simply to add?

Turn now to the opening of the orchestral score and note the following:

  • m.1: addition of a new quarter note part which is distributed over the trumpets on the one hand and the rest of the orchestra on the other, thus giving rhythmic life to a ‘mere’ sucession of chords on the beats. (See the Cruxifixus from the b minor mass for another example of Bach using this device);
  • m.2: orchestral parts join in the solo parts’ repeat of the opening rhythm. Note that they start on the second 8th notes of the measure, reflecting the original rhythm and jumping on the moving train of 16th notes in that solo part;
  • m.4: a similar strategy, doubling the repeat of the solo line melodic idea, this time on the second descent from D, and again from the exact position (second 16th note of the beat this time);
  • m.9–12 following: bringing out the melodic 8th note motion with parallel 3rds above and below, alternating by the measures (m.9,11 above, m.10, 12 below). Note how m.10 and 12 also resemble the opening rhythm, if not its contour;
  • mm.14–16: a similar process, all below, but alternating 3rds and 6ths (effectively ‘swapping’ F# and D between solo and orchestra).
  • m.(13,)16: the pedal D in the solo line (m.13) moves into a separate bass pedal (16).
  • m.20: orchestral chords verticalize the cycle of fifth harmonies and reduce the orchestra’s metrical levels to measure-only.

Overall, note:

  • how Bach has not only picked out implied lines, but been relatively free in the addition of germane material to thicken out the orchestral version. This does not imply any deficiency in the original; it’s Bach’s way of realizing the same material for different media;
  • the key change from E for the violin to D for the orchestra (in connection with our discussion of the Chopin example above);
  • how consistent this piece is with Lester’s (1986, p.138) observation that Bach will often project multiple metrical levels simultaneously and explicitly: here we have very explicit pulses at 16th, 8th, and 1/4 note levels for most of the time.

Mussorgsky: ‘Gnomus’ from Pictures at an Exhibition. (orch. Ravel and Stokowski)

Pictures at an Exhibition is one of those works that orchestrators can’t seem to resist so it seems fitting to use that as the case-study to ask:

Question: ‘How have composers handled the same material differently?’

Let’s look as the ‘Gnomus’ movement as an example. The picture in this case depicts a sketch for a toy nutcracker shaped as a deformed gnome; the music is accordingly fragmentary and ghoulish. Again, start with the piano original and come up with your own ideas for how to orchestrate it; we’ll then go through each successive piece of material in the movement as realized by Mussorgsky (M), Ravel (R), and Stokowski (S). Mussorgsky and Ravel are on IMSLP; regrettably the Stokowski is not.

m.1: We start with a dramatic run of 8th notes across two registers.

  • M slurs by the bar and changes the dynamics on repeat ( ff /p ).
  • R follows the same slurs as M and uses the same low strings and wind timbre for both ff and versions.
  • S opts for multiple articulations of bar 1 (note the wind slurs by register and the strings’ accents). Unlike R., S. changes the orchestration for the m.3 p.
  • Note how neither R nor M separates the part out into separate registers. The difficulty of the disjunct line works well for this pictorial effect.

m.2, 5: Pause note

  • M: Note the notated sfzand remember the natural diminuendo of a piano chord’s decay.
  • R: The existing parts provide the attack; two horns provide the sustain with explicitly notated diminuendo. Note that the horns are muted the 2ndtime for the quieter version, and for timbral variety;
  • S: Tutti on pause notes; no diminuendo.

m.8-9: Hemiola

  • M: indicated by sfzs;
  • R and S: also use accents and some secondary (incomplete) parts.

m.10: Treble flourish

  • M: dispenses even with accents in this case;
  • R and S: add percussion to the downbeat. (NB: S does not include the second phrase but proceeds straight to m.17).

m.19: At the double bar

  • M: high treble with accents on both beats.
  • R (reh.8): high winds with pizzicato for attacks (on the 1st beat, not the 2nd). Tuba theme (with a little from the trumpet on the last note, likely for both practical reasons (allowing the tuba to breathe!) and perhaps for a timbral sleight of hand not unlike the horns at the beginning. The second time (reh.9) sees the ethereal addition of harp and celesta on the main, accented material and of a string glissando that did not (almost could not) exist in the piano version;
  • S (reh.1): string pizzicato on beat 1 versus trumpets and accents on second beatwhich add an antiphonal effect to the the passage;

m.36-7: Before the meno mosso

  • R: subito p
  • S: crescendo (sic, not diminuendo as in M.) and the addition of timpani to the GP bar, (then diminuendo).

m.38: Meno mosso

  • M: slow parallel writing with the low Eb as a kind of pedal inviting separation
  • R and S: indeed allocate some part to the Eb pedal only.
  • S (fig.2): adds hairpins peaking at the half-bar.


  • M: Loud, sequential descending passage
  • R (fig.14): full wind plus trumpets 2 and 3; string glissandos on the lower part (which is now becoming a characteristic gesture of this orchestration).
  • S (fig.5): flute, piccolo, horn, trumpet; strings tremolo but not glissando. The allocation of instrumental sections changes at fig 6.


  • M: Low murmuring; treble upper part version of earlier material.
  • R (fig.15): Alternating bass clarinet and bassoon. Double basses and cellos share the note and glissando respectively. In the treble, we have flute, violin, and harp, a choice which could be seen as a combination of earlier versions.
  • S (fig.7): full, low wind and bass drum, all playing throughout. Treble parts as previously (with the same swap).

m.94: Big flourishing finish

  • M: a spare treatment, with just two voices until the final chord.
  • R (fig. 18): addition of voices (note how those voices relate to the original pitches and which metrical positions they start on);
  • S: adds in a longer version of the Cb, perhaps mindful of the other moments in which this movement has focused on the single, sustained pitch. All parts additions begin on downbeats, with the primary additions on the penultimate measure.

In general, Stokowski adopts an approach that:

  • Closely resembles Ravel’s more distinctive additions (including even eg. the glissando additions … );
  • More freely changes the material (in cutting all repeats, for instance);
  • Tends to joins up successive sections somewhat (sustaining through rests, for instance), perhaps seeking to avoid this short work becoming too fragmentary.

Stravinsky: ‘Danse Sacrale’ from the Rite of Spring (bass part). (orch. Stravinky)

On now to another iconic piece of music that has a rich concert life in both piano and orchestral forms. We also return to:

  • bass line considerations (and to a bass line that consists of few pitches);
  • the orchestral crescendo which has featured through this chapter and section.

Here the material consists of strictly separate elements that are rhythmically re-worked in a dizzying range of ways. Fundamentally, however, we have a simple alternation between a bass part on the beats and a treble response off-beat. The treble part takes a few different forms, mostly centred on short, sharp, accented chords, but also intermittently including the classic Stravinsky crescendo: a short, rapid crescendo from a metrically weak position to a strong one. This is not indicated in the piano version (it’s not possible there for sustained chords) but is a highly distinctive feature of Stravinsky’s orchestral writing, here with horns and trombones for the crescendo and more brass added to mark the down beat.

Here we’re going to focus on longer range orchestral crescendo the bass and ask:

Question: How would you score the bass in the main, ‘A’ sections (of this ‘rondo’) as part of a gradual orchestral crescendo?

Like many, I first encountered this music aurally, (and was blown away by it, of course!). Later, when I came to check out the score, the first thing that struck me was the bass line. I remembered the colossal forces involved, and the alternation between bass and treble, and I suppose I just assumed that the orchestration involved some kind of parity between the forces allocated to bass and treble. Instead, I found the bass very lightly scored (at least initially) and gradually changing throughout, creating timbrally variety and a gradual crescendo both within sections (especially in the passage from reh. 186) and between them. Here’s a brief summary of the process first entrances:

  • Rehearsal Mark 142 ff. (start): double basses and timpani, initially alternating (i.e. only one at a time!) then briefly together on the first octave downbeat at reh.144. The tuba is then added, for a new alternation double bass and timpani alternation, the tuba siding with the basses (and the lower octave) until 146 where the double basses and timpani are re-united (no tuba) for the more melodic D-F alternation.
  • Rehearsal Mark 167 ff. (c.2’00’’ in the recording below) sees bassoon and contra bassoon initially take the place of the tuba, and then alternate with it in the same pattern as before. Double basses and timpani remain.
  • Rehearsal Mark 180ff. (c.3’00’’): a brief snatch of the A section here sees no bassoons, but tuba, double basses, timpani and the introduction of the bass drum before the final section …
  • Rehearsal Mark 186ff. (c.3’40”): This final section initially start with all the action concentrated in the bass register (i.e. both ‘bass’ and ‘treble’ of before). Bassoons and tubas split, covering both the ‘treble’ and ‘bass’ between their members, so we’re now more in the territory of self-contained subsections (rule of thumb from the first chapter) than the more pointillistic approaches (see this chapter) we’ve seen so far. The double basses and bass drum are in, but the timpani is removed for the first time, returning only when the true treble range recommences at rehearsal mark 189. Rehearsal mark 190 sees the introduction bass clarinet, then two very low horn (VI and VIII) at reh.192. Rehearsal mark 195 finally sees the true tutti with the bass clarinet, 3rd bassoon and contra bassoons, horns VI and VIII, timpani, bass drum, and double basses, but not tubas – reallocated to the heavy brass for the ‘treble’ off beats. 
  • the end of all of this variety, the single, unequivocal sfff bass cluster chord seems an especially fitting end suitably prepared with equally garish flourishes in a genuinely treble register. Even here, the final downbeat does not feature all the bass instruments, but adds cellos for the first time and omits those of the wind section.

Bartók: Eight Improvisations on Hungarian Peasant Songs (orch. Gotham)

I’ll round out the chapter and section by putting my head on the block with a full orchestration of my own. If nothing else, this should help to avoid leaving the impression that only the likes of Bach, Ravel, and Gubaidulina can do this kind of thing (brilliant though they doubtless are).

Here is Bartók’s own piano part; you can find my orchestration from the same page under the ‘arrangements and transcriptions’ tab.

I began this process with extreme attention to Bartók’s fascinating articulations and found more about what he means by them through engaging with his other scores, and by chancing upon his own glossary of articulations as used in his 1916 edition of J.S. Bach’s Anna Magdalena Notebook (see the title pages). In orchestrating this work, apart from realizing the implications of the score (highly inviting to orchestrators!), I also sought to realize something of Bartók’s own style.

Here are more specific comments for the successive decisions.

  • m.1: increase the sound and density of the opening flourish by the addition of octaves, by introducing several simultaneous rhythmic modifications of the triplet, and by inviting the double basses to play a snap (a.k.a. ‘Bartók’) pizz on their lowest string, preferable on the ‘correct’ pitch C0 (for basses with extensions), but if not, then still preferring the open string to shifting the pitch up the octave.
  • m.2: for the repeated chords (with the piano’s pedal down), alter the octave to make use of the four horns in a favorable range and to outline the outer voices of that chord with a battery of open strings.
  • m.5: where the piano part sees doubled octaves, the orchestral version attracts concomitantly more.
  • m.6: suddenly, the piano pedal comes off so that’s a cue to thin the texture (removing the added bass, for instance). The trumpets serve to mark that moment and contribute to the gradual move from repeated off-beat chords to the more active figure and emerging secondary line.
  • m.9: ‘con grazia’ in this range and in this slightly rustic context calls for the oboe.
  • m.11: suddenly the dynamics, texture, and tessiture change again. The glissando is another Bartók hallmark and, in combination with the high e harmonic, adds a bravura flourish.
  • m.13: the alternation of two pitches in the bass register a 4thor 5thapart is very suggestive of the timpani.
  • m.19–20: here’s a chance to flex out dovetailing muscles as part of covering a large range with crescendo.
  • m.21: reserve something (percussion) for final sffchord. The timpani flourish is a creative addition which brings that soloistic part to a climactic close along with the wider section.
  • m.22: ‘leggiero’ is the main cue here. Use like timbres for the clusters. Oboes fit the leggiero bill here and flutes round the sound out a little.
  • m.23: The accent invites a little creative change: I opt for a solitary string pizz and sending flute up one octave for this note only.
  • m.24: Cor anglaise and solo cello duet sonority begins before the ‘true’ duet from m.28. (VC (moving from oboe to VC-Cor 28 duet). Solo / altri exchange.
  • m.26: The equivalent accent to m.23 is similar, but less directly implied by the piano part (deduced from the solo line).
  • m.28: The articulation in the cello melody here is a little fussy, but should help to bring out the section’s ‘capriciousness’.
  • m.29: Yet another reworking of the accompaniment accent, here as a slightly different violin figure (similar to both m.23,26 and the s.drum in m.13).
  • m.35: The texture (and register) develop here, notable expanding the use of the woodwinds.
  • m.43: Bassoon and contra bassoon share the octave leaps of the original; double basses connect the two somewhat with a semi-tone version.
  • m.46: The higher octave here allows us to return to a timbre and gesture from earlier on: the ‘throw-away’ high e harmonic on the violins.
  • m.48: Once again, the bass ‘line’ is separated between the F pedal and the moving part.
  • m.51: A variety of solutions to the clear ‘attack-sustain’ gesture here. The solo timbres are as before, lending some continuity to avoid too excessively fragmentary a feel.
  • m.53: This two voice-invention with occasionally double octaves is an orchestrator’s dream. Trombones take centre stage, horns pick out the quarter note exchange as a kind of ‘bell peal’ which is joined on every other measure by tubular bell (suitably enough!).
  • m.63: Changes to the horn parts bring out the rhythmic acceleration.
  • m.65: Ruomoso? Al talone? All down? The Rite of Spring looms large.
  • m.67: Subito mf reduced the strings to pizz. But introduces winds in preparation for the crescendo.
  • m.69: The big tune returns! Having flirted with a Rite of Spring esque all down passage before, it seems fitting to try the legato version here. Sound covered by brass and off beats.
  • m.77: Note the move into a higher octave here (in both versions).
  • m.81: The sextuplet requires some re-working for orchestra: here some parts play all 6 notes, but in close position and octave doublings; other parts (generally the lower ones) pick out a triplet version. The snare drum continues this 6tuplet through the downbeat (an addition to the original) up to the final chord.

  1. For instance see Rimsky-Korsakov's Preface to his iconic Principles of Orchestration, 1891.


Icon for the Creative Commons Attribution-ShareAlike 4.0 International License

Open Music Theory by Mark Gotham is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License, except where otherwise noted.

Share This Book