Laying down the backing tracks

Once you have finished creating your Melodyne arrangement and the ink (so to speak) is dry on the blueprint, it is the turn of the singer to lay down the various backing tracks voice by voice. Nothing turns on the order in which these are recorded, so let him call the shots.

The headphone mix

Up front in the headphone mix should be Melodyne’s blueprint version of the track he’s laying down. Also present, but noticeably quieter, should be the blueprint of the other voices – loud enough to supply the harmonic context but not to distract him from the voice he is supposed to sing and perhaps cause him to switch voices accidentally in mid track. Inexperienced singers are particularly prone to muddle up the various voices in this way. That is why it is so useful to have a Melodyne blueprint as an audible point of reference: It helps the singer stick to the right voice.

Double-track the harmonies at the same time

Once the singer has mastered each voice, he should lay down three or four identical takes to use as doubles. A better occasion to do this is unlikely to present itself. The essential points to remember when doubling existing tracks have been described elsewhere.

Register problems

It may be that your arrangement strays in places beyond the vocal range of the singer or requires him to sing in a register in which he is uncomfortable. The second voice may only be a third or a fourth above the lead and perfectly manageable, but the third voice could at times be an octave higher, or even more, and exceed his technical capabilities.

The usual solution is to bring in additional singers. It’s generally no problem for a female vocalist to double a male one at the octave. Conversely, if the lead vocalist is a woman and the backing vocals are lower in pitch, she might prefer to leave them to a man.

Falsetto

If a singer has to resort to falsetto to sing one of the higher voices, that in itself is not a problem – even though the results may sound a little thin heard in isolation and perhaps not in keeping with the temper of the song. It is what the harmony vocals contribute to the ensemble that matters, not how they sound on their own, and their purpose in this case is to enrich the tone color of the lead vocal track and improve the overall texture. The difference between the singer’s falsetto timbre and that of his normal voice is rather to be welcomed, as it extends the palette of available tone colors.

Using multiple vocalists

The use of different singers to supply the harmony vocals might also be desirable artistically: The effect you are trying to create might be that of a choir or backing group. If that is not the case, however, the use of other singers could limit your creative options from the standpoint of sound design, because however low the second vocalist may be in the mix, he or she will always be perceived as a separate performer. This is not the case if the parts are all performed by the same vocalist, where the illusion of a single take of a single voice – but one with an interesting tone color – can be preserved.

Artificial extension of the vocal range

If you only have one singer to work with and the voice leading you have decided upon takes him outside his range, the following trick should help:

Instead of the (too high) third voice, ask him for an extra take of the (manageable) second voice. Then shift each in turn of the resulting notes upwards in Melodyne by the requisite interval. In this way, you can obtain the third voice simply by transposing your (spare) recording of the second. Melodyne studio’s multitrack display helps here:

The “target voice” (which is outside the singer’s vocal range) is already present in the Melodyne blueprint – in this case, on the track “layout hi”. Switch this track to “reference” and it will appear as gray blobs in the display.
Your raw material is your spare take of the second voice (which was within the singer’s vocal range); this is “take 2” in the track list and you switch this to “edit” so that its blobs appear in orange in the display. All you need do now is shift the orange blobs upwards, one by one, until they are superimposed upon (i.e. congruent with) the gray ones.
The result might look like this:

The only places where this congruence is undesirable is where the blobs represent breaths or sibilants; these should not be shifted upwards as they sound more natural at their original frequencies.

After creating a part in this way by pitch-shifting, you should experiment with the formants to see if this gives you a more natural sound. Experience shows that a slight (ca. 10-cent) formant shift downwards works well with voices that have been transposed upwards. Conversely, if a voice has been transposed downwards, you should shift the formants up a little.
In our example, a slight adjustment to the formants of certain notes is all that was required: