Music Hub › Guides › Theory and harmony › Singing a vocal harmony for two or three voices

A vocal harmony is a second line that follows the chord, not the melody alone

Singing a harmony, in the sense this guide gives the word, means adding a second voice that moves along with the melody, note against note or nearly so, but at a different pitch at every moment. This isn't polyphony in the scholarly sense of the term, where two independent melodic lines cross freely, each following its own shape: in a song for two or three voices, the second line follows the rhythm and overall shape of the melody, hugging its phrases, but sings a different note of the current chord. That's a useful distinction to keep in mind even before talking about intervals: a vocal harmony works like an accompaniment part that happens to be sung rather than played, somewhat like a pianist's left hand laying down a chord under a melody, except that here each singer carries only one note at a time.

The vocabulary of group singing has long had a name for this practice: close harmony, two or more voices that stay near each other, most often less than an octave apart, mostly in thirds and sixths. That closeness isn't an aesthetic accident: it corresponds to the zone where two voices blend best to the ear without one masking the other, and it explains why almost every tradition of multi-voice singing, from gospel to barbershop, is built on these two intervals rather than on wider gaps like the fifth or the octave, which are reserved for more specific uses detailed later in this guide.

So what the second voice sings is never arbitrary: at every moment, it has to match a note belonging to the current chord or scale, or what was meant to sound like a harmony sounds like a wrong note instead. It's this constant link to the chord of the moment, rather than to the melody alone, that structures the rest of this guide: how to choose that note, how it changes from one chord to the next, and what a possible third voice does once the first two are in place.

Third or sixth, above or below: the two reference intervals

The two intervals that come up most often when building a vocal harmony are the third and the sixth, sung indifferently above or below the melody. A third below remains the most spontaneous option for most singers, probably because it matches the first reflex adopted when accompanying someone else by ear; a third above, a sixth below or a sixth above each give a slightly different colour, brighter when the second voice climbs, more enveloping when it descends.

This choice between above and below isn't just a matter of taste: it also depends on the comfortable tessitura of the singer holding the second voice, a subject covered in its own right in the Music Hub guide on finding your singing key. The same melody can be harmonised in a third above for a voice that's comfortable in the high end of the passage, and in a third below for another, lower voice on the same song: the chosen interval changes nothing about the harmonic logic as long as it stays a third or a sixth, only the direction changes.

The sixth is really nothing more than a third turned upside down at the octave, which makes the two intervals much closer to each other than they first appear. A minor third below a note and a major sixth above that same note point to exactly the same destination note, only placed an octave higher; and the quality of the interval flips mechanically in the process, a minor third becoming a major sixth, a major third becoming a minor sixth. Knowing how to recognise your third is therefore enough, in practice, to know your sixth without having to memorise a second table: all it takes is turning the same note up an octave and swapping major for minor or the other way round.

This table builds the diatonic thirds on each of the seven degrees of a major scale, illustrated here in C major: it already shows, even before this guide gets to the theoretical reason behind it in the next section, that the quality of the interval isn't constant from one degree to another. A third below the first degree is minor, while a third below the third degree is major: the same distance, called a third either way, actually covers two different gaps in semitones, three or four, depending on the starting note.

Why the interval changes along the way: staying inside the current chord and scale

The most tempting reflex for anyone starting out in vocal harmony is to find a good third on the song's first note, then reproduce the same gap in semitones on every note that follows, as if the harmony had to stay at a fixed chromatic distance from the melody from start to finish. That reflex never survives the test of the ear for long, because a scale isn't symmetrical: its tones and semitones don't repeat at regular intervals, which the table in the previous section already shows. A third below the tonic isn't the same size in semitones as a third below the next degree; keeping the same chromatic gap on both therefore means stepping outside the scale on one of the two.

The right rule runs the other way: it isn't the gap in semitones that stays fixed, it's the position within the scale and the chord that stays fixed — in this case a third or a sixth further down or up the staff. Concretely, if the melody holds for a moment on the tonic, the scale's first degree, a harmony a third below lands on the sixth degree, a minor interval. If the melody then moves to that sixth degree itself, that same rule, a third below, this time lands on the fourth degree, still within the same scale, but this time with a major interval, a different gap in semitones from the one before. A singer who has mechanically kept the same number of semitones instead of recalculating their position in the scale ends up, on this second chord, on a note foreign to both the chord and the key.

This constraint isn't unique to singing: it's exactly what any accompaniment that follows a chord chart does, whether vocal harmony or a chord voicing on a keyboard, and it explains why a good ear for vocal harmony looks a lot like a good ear for chords themselves. The Music Hub guide on finding the chords of a song by ear details, from the accompaniment side, this same reflex of staying within the current scale rather than guessing at random: here, it applies to a single sung voice rather than to a whole chord laid down on an instrument, but the underlying logic doesn't change.

Choosing the right note in the current chord: the method, step by step

In practice, choosing the harmony note on a given chord happens in three short steps, repeated for every new chord in the chart rather than done once for the whole song. The first is to spot which note of the chord the melody occupies at that exact moment: its root, its third or its fifth. The second is to count, within the scale of the current key, a third or a sixth higher or lower than that melody note. The third, the one a hurried singer most often forgets, is to check that the note found this way really belongs to the chord sounding at that moment, and not just to the song's general scale: the two coincide most of the time, but not always, particularly on a chord that borrows a colour from outside the base key.

This last check matters more on long notes, the ones that hold for a good part of a bar and carry most of a passage's harmonic colour, than on a quick passing note, slipped between two more stable points: on those, the ear is quite happy to accept a harmony that simply follows the melodic movement without stopping to check every note against the chord of the moment. This hierarchy, lingering on the notes that carry the chord and staying looser on the ones that merely pass through it, looks a lot like the one a guitarist or a pianist already applies to accompany a chord chart, and it's no exact science reserved only for instrumentalists.

One shortcut helps speed up this check without having to name each interval by counting semitones: the third and the sixth, applied correctly, almost always land on a note of the current triad, because a triad itself is nothing more than a stack of thirds. If the melody sings the root of a chord, the neighbouring third in the scale very naturally lands on that same chord's third; if the melody sings the chord's third, the next neighbouring third lands on its fifth. That's no coincidence: it's the direct consequence of every chord's structure of stacked thirds, and it's what makes the diatonic-third rule as reliable as it turns out to be in practice.

The third voice: completing the triad

When a second harmony is added to the melody and a first voice already set a third or a sixth away, its most common function is no longer to choose between a third and a sixth for itself, but to complete whatever the chord is still missing. A complete triad has three notes, the root, the third and the fifth; if the melody and the first harmony already cover two of those three notes, the third voice's job is to go and fetch the one still left silent, rather than reflexively doubling one of the two already present.

This division of labour explains why a vocal trio generally sounds fuller than a duo, not simply because there's one more voice, but because the chord's three distinct notes can finally sound together at the same moment. A duo, by contrast, always leaves one note of the triad merely suggested to the ear, absent from the actual mix but mentally reconstructed from the two notes that really do sound: that's enough for most purposes, but it isn't the same harmonic density as a trio covering all three notes at once.

This completion logic isn't unique to two- or three-voice singing: in four-singer formations like the barbershop quartet covered later in this guide, it extends further, adding a fourth note beyond the plain triad, most often a seventh that enriches the chord even more. The principle stays the same every time a voice is added: look for the chord note still missing from the mix rather than reinforcing one that's already there. For a trio, sticking to the plain triad, root, third and fifth, is more than enough to get a full, recognisable chord, with no need to go looking for a fourth colour.

Octave doubling: a different option from a true harmony

Two singers who sing exactly the same melody note, one an octave higher or lower than the other, aren't producing a harmony in the sense this guide has defined so far: they're singing the same note, at the same moment, in the same function relative to the chord, only transposed a full octave. No new chord note appears in the mix: octave doubling changes the register and adds power or roundness, but it doesn't add any extra harmonic colour, unlike a third or a sixth, which do bring out a different note of the chord.

That doesn't make it a technique to dismiss, though: it serves a genuinely different purpose very well, reinforcing a melody rather than harmonising it, particularly on a chorus meant to sound bigger or more carried without changing its colour. The Sacred Harp tradition, that American choral singing style born from the New England singing schools at the end of the eighteenth century and fixed in its reference tunebook in 1844, illustrates this almost in spite of itself: its vocal parts are sung a cappella, in a hollow-square formation, but two of them, the melody part and the one that answers it, are each held by both men's and women's voices an octave apart. A harmony written for four parts ends up sounding like six distinct voices this way, without a single extra chord note ever being added: the richness comes from octave doubling, not from a fifth or sixth harmonic note.

The distinction worth remembering, then, has less to do with the quality of the effect, which is often very successful, than with its nature: octave doubling reinforces a note that's already there, a harmony in thirds or sixths adds a new one. Confusing the two leads to thinking a passage has been harmonised when it's only been thickened, which isn't a flaw in itself but never replaces, to the ear, the colour of a genuine second voice.

Voice crossing: why it's generally avoided

A harmony placed below the melody stays, most of the time, below it for the whole length of a phrase; a harmony placed above stays above. Voice crossing describes the moment when that position flips without warning, a voice meant to stay low momentarily climbing above the melody it's supposed to be accompanying, or the other way round. The general rule in multi-voice singing is to avoid this crossing rather than seek it out, for a reason that has less to do with theory than with how the ear tracks one line among several.

The ear picks out a lead, a principal voice, largely thanks to its relative position among the other voices sounding at the same time: the highest line, or the most central one depending on the arrangement, is spontaneously perceived as the one carrying the meaning. An unprepared crossing blurs that cue: the moment the voice meant to stay below climbs above, the ear gets the impression, for an instant, that the two voices have swapped roles, which throws off the listening even though each note, taken on its own, stays perfectly in tune. On stage, that same crossing also complicates the singers' own work, since they rely heavily on their voice's position relative to their partner's to judge whether their own pitch is still holding.

That doesn't make crossing a mistake under all circumstances: some arrangements use it deliberately, at a climactic point or the end of a phrase, precisely to create an effect of surprise or tension before returning to the usual position. The difference lies in intention and preparation: a crossing decided on and rehearsed produces an effect; an accidental crossing, caused by a poorly set tessitura or a hesitation live on stage, produces only confusion. In general, for a duo or trio just starting out with vocal harmony, keeping each voice on its own side of the melody the whole way through remains the safest choice, and the simplest to hold on to in concert.

Parallel fifths and octaves: banned in classical writing, common elsewhere

Classical four-voice writing, the kind European harmony treatises have taught for centuries, bans one specific movement between two voices: the parallel fifth and the parallel octave, meaning two voices that move together, in the same direction, keeping the same gap of a perfect fifth or a full octave the whole way through. This convention isn't recent: the systematic avoidance of parallel perfect consonances settles into compositional practice around 1450, just as Renaissance polyphonic writing moves away from the earlier medieval practices that, by contrast, made regular use of them. It was then codified in the teaching of so-called species counterpoint from the eighteenth century onward, and remains to this day the first rule anyone learning choral-style four-voice writing is taught.

The theoretical reason has to do with the independence of the voices: a parallel fifth or octave makes two voices sound as though they were really just one, doubled, because the ear has trouble perceiving two lines moving in a rigorously identical way. What classical four-voice writing seeks to preserve is exactly the opposite: voices that each keep their own motion, meeting on common notes or moving in contrary motion rather than in strict parallel on a perfect interval.

This rule, though, is nothing like a universal law of music: it describes a goal specific to one particular style, choral writing inherited from the European classical tradition, not a ban that would hold for all music written or sung in multiple voices. Whole swathes of rock, pop and a good deal of gospel instead actively seek out that same parallel motion, fifths and octaves included, as a colour in its own right rather than a flaw to fix: the electric guitar power chord, which is nothing more than a parallel fifth played continuously, is the most direct example, and a choir that moves a whole block of chords in parallel motion is going after exactly that compact, massed effect that classical writing was trying to avoid. The right reflex, then, isn't to ban this motion under all circumstances, but to know what style you're writing in before deciding whether it serves the music or works against it.

Gospel and bluegrass: harmonising by ear, in the moment

Some multi-voice singing traditions were built on direct improvisation rather than on an arrangement written out in advance, which makes them an excellent place to watch the method described earlier in this guide applied in real time, by singers who have often received no formal theoretical training at all.

African American gospel is the most direct example, through the jubilee quartet tradition, which traces back to the Fisk Jubilee Singers, an ensemble organised in 1871 by George L. White to make the Negro spirituals known beyond the American South. Quartets then formed in churches, colleges and even factories from the early twentieth century on, specialising in close harmony sung a cappella, often built around call-and-response: a soloist launches a phrase, the group takes it up and completes it in harmony, with no sheet music, each singer fitting their note by ear against the chord the others are already holding. A group like the Soul Stirrers evolved this style from the mid-1930s onward, moving toward a more modern gospel quartet sound, blending the precision of jubilee-style close harmony with a far more improvised fervour in the fine detail of the voices.

Bluegrass, for its part, owes the bulk of its vocal convention to a single musician, Bill Monroe, who pushed his own tenor voice into a register unusually high for the time, up to pitches few singers dared aim for, to blend it into a three-voice harmony that has remained the style's signature ever since — what came to be called the high lonesome sound. His band's 1945 lineup, the Blue Grass Boys, with Earl Scruggs on banjo, Lester Flatt on guitar, Chubby Wise on fiddle and Howard Watts on upright bass, is regarded today as the genre's founding moment. In bluegrass vocabulary as in gospel quartet vocabulary, the voice called tenor isn't necessarily the highest in the sense of an individual voice type, either: it names the part that sings above the lead in the arrangement, a naming convention for the part rather than a vocal category, and this assignment is generally decided by ear, during rehearsals, rather than on a staff written out in advance.

Barbershop, doo-wop and Sacred Harp: harmonies handed down and rehearsed

Other traditions, by contrast, have sought to fix their harmonies in place, through a notation system of their own or through arrangement work rehearsed until every voice is learned by heart, rather than reinventing them at every performance.

Barbershop organised itself into a structured movement starting in 1938, against a backdrop of nineteenth-century African American vocal tradition made up of hymns and popular four-voice songs; the genre's history records that a split then took place, with part of the style settling into a form popularised by this organised movement, while jazz and vocal gospel kept evolving separately elsewhere. Barbershop's harmonic signature remains, to this day, a particular seventh chord, in which the seventh is tuned slightly lower than on a tempered keyboard, so as to make the four voices' harmonics line up into what's called a ringing chord, where the frequencies add up almost like a fifth instrument. Arrangers in the discipline aim for a high proportion of this type of chord, often more than one chord in three across a song, precisely because it's this particular seventh that gives the style its colour, recognisable above all others.

Doo-wop, which emerged from the street vocal groups of the 1950s, arranges its voices in an almost identical way from one song to the next: a tenor lead carries the melody, a second tenor voice and a baritone voice blend into a single intermediate harmonic mass, a bass voice holds the root down low and sometimes opens the song or punctuates its phrases, and a falsetto voice occasionally tops the whole thing off near the end. Groups like the Ravens or the Five Keys, right at the start of the 1950s, replaced the plain vocal drone with more rhythmic, breathy syllables, a technique that eventually became one of the genre's trademarks.

Sacred Harp, already met earlier in this guide in connection with octave doubling, inherits from the singing schools that appeared in New England between 1770 and 1820 and were fixed in its reference tunebook in 1844: notes printed in distinct shapes, triangle, round, square and diamond, let singers read their part without any classical musical training. Sung a cappella, in a hollow-square formation where each vocal section faces the other three, this repertoire remains one of the oldest documented examples of harmonised multi-voice singing passed down almost without interruption for two centuries.

These five traditions don't all share the same history or the same relationship to written notation, but they converge on one point: the colour of a harmony has less to do with the number of voices than with how precisely each one holds its note of the chord, whether that note is found by ear as in gospel and bluegrass, or handed down from generation to generation as in barbershop, doo-wop and Sacred Harp.

Assigning who sings what by each singer's voice type

In a duo, the question of who sings the melody and who sings the harmony, and whether that harmony sits above or below, is most often settled by each singer's tessitura rather than by personal preference. Classical singing names these tessituras soprano, alto, tenor and bass, from highest to lowest; those words are enough here to place who, in a duo or trio, naturally takes the highest or lowest part of an arrangement, without needing to go any further into the detail of these classifications.

This assignment doesn't, however, come down simply to comparing who sings highest in absolute terms: it mainly depends on the zone where each voice stays comfortable across the whole song, chorus included, not just on its single most flattering passage. A harmony that pushes a singer into an uncomfortable zone for several minutes tires the voice and loses accuracy, even if that singer's voice can, on paper, reach that pitch briefly. Precisely measuring that comfort zone, calling it tessitura rather than range, and working out how far to shift it for a given song is the entire subject of the Music Hub guide on finding your singing key: this guide stops at the question of assigning parts between singers, not the individual measurement of each voice.

In a trio, this assignment gets slightly more complicated, because three voices have to spread across three notes of the triad without two of them accidentally landing in the same register to the point of getting in each other's way. A common practice is to place the melody in a middle zone and hand the two harmonies to the voices that feel most comfortable above and below that zone respectively, rather than imposing a fixed register on each singer regardless of the song being played. This flexibility, decided case by case rather than fixed once and for all, explains why the same trio can swap parts from one song to the next depending on the chosen key, without that calling each singer's voice into question in absolute terms.

The reflex worth building: holding your line while the other voice moves

The most common difficulty for anyone discovering vocal harmony isn't finding the right note the first time: it's keeping it once the other voice, the melody or a second harmony, starts to move. The ear naturally tends to follow the most familiar or loudest line in the mix, which, for a beginning singer, often means falling back onto the melody itself without even noticing as soon as it changes note, instead of staying on their own path through the chord.

This reflex of independence is trained separately from pitch accuracy itself: it isn't enough to know, in theory, which note to sing, you also have to be able to hold it without getting pulled along by what the other voice is doing. The most effective exercise is to first isolate your own line against a fixed reference, an instrument playing the melody or a recording, before singing it alongside another singer live: hearing your note resist the melody's change, without being exposed to real live vocal interaction yet, anchors the line far faster than a direct two-voice attempt would.

A complementary exercise, common among group singers, is to hold a harmony note while a partner or a recording deliberately moves the melody all around it, without changing your own note, then to swap roles: this back-and-forth game trains you to perceive the melody as something external to your own line rather than as a magnet pulling your voice toward it. With practice, this reflex stops being a conscious effort: the ear learns to hold its position within the chord the same way it holds its position in a conversation, listening to the other person without repeating what they say.

Common pitfalls to avoid

Explore in the Music Hub

More guides