Friday, December 27, 2013

Song festival - the producer speaks

I wrote last month about the song festival which is going to be held on the kibbutz at the end of December; I wrote the music for two songs, as well as arranging them and recording demos. I also arranged the music for another song, which already had a tune. The organisers decided that the lyricist of the first song would sing her song at the festival, but that it would be better to record a vocal track 'in the studio' first, as she's not too good a singer.

So this lady came round to my home; I put on my producer's hat and started talking to her about how the vocal should be sung and how it would be recorded (making this the first decision to be made about which microphone should I use - is she going to sing softly or loud?). We also had to chose a key: I had originally recorded the song in Em, but had subsequently lowered it to Dm. After trying some keys, we eventually settled on Bm, this being the most comfortable for her. Once we had the key sorted, I could create a new music track. I then recorded several takes of her singing, so that I would be able to produce a good composite vocal.

After she went home, I started making a composite track of the vocals. I listened to the takes and decided which take would contribute each part. In the end, the verses came from one take and the chorus (duplicated) from another. As the second chorus is a bit complicated - two lines in the original key then the entire chorus in a higher key, I had to perform a fair amount of digital trickery in order to get everything right. I used equalisation in order to take out a harsh note in her voice, and reverb made the vocal sound softer.

Once I had a complete vocal, I listened to it along with the music. I was disappointed to discover that her timing was off on some lines; whilst it's possible to fix this by chopping milliseconds out of the vocal track (and adding them back later), I don't like doing this very much, and normally it's easier to rerecord the vocal. So the next day, I got the lady back into the studio and she sang the first line only several times. I then replaced the original first line with the best take of these edit pieces, thus improving the entire vocal.

I had previously recorded myself singing the song in Dm; I used the pitch changing software to lower this recording to Bm, thus my voice was singing at what is for me a very low pitch. I edited this recording to leave only the chori (including the modulation) and mixed this low in the background, thus adding 'oomph' to the chorus.

The singer was knocked out by the final version, but I think that this is a natural reaction to someone hearing their words realised as a song. I warned her not to think that she sings as well as she sounds in the final recording. In the live performance, she will either sing very quietly, have her microphone turned off (shades of Robbie Robertson) or simply mime.

The organisers had a much bigger problem with the second song as they couldn't find anyone to sing it. After discussion, it was decided that they would play my recording of the song (ie with me singing) and accompany it with slides projected on a screen. Once this decision had been made, I realised that I would have to improve my vocal as originally it had been intended only as a demo - and at the time of recording, I too wasn't familiar with the tune and had problems recording it. As my wife pointed out that I was mispronouncing a word which appeared in each of the first two verses, I decided to rerecord the first two verses.

The first verse is 'just me' - one voice with a little reverb. For the second verse, I duplicated the vocal track, made the track stereo, added a little delay to one channel, then mixed this track into the song with more reverb and more bass in the eq. The result isn't exactly what I wanted, but it's good enough: there is a subtle difference (more "depth") between the second verse vocals and the first verse. The third verse, though, is something different ... as opposed to the first song, in which I apparently sing in a very low voice, this song has me singing harmony to the final verse in a very high voice, effortlessly reaching E an octave above middle C.

I hadn't any idea when I started of what the harmony should actually be; putting the harmony a third above the tune worked for most of the lines but not all. I avoided parallel fifths and octaves and ended up with something that sounds quite reasonable. One has to remember that this harmony vocal is mixed behind the lead vocal so it isn't heard on its own. At one point, I got slightly confused as to which chord was being played in which bar; as a result, the harmony for that bar creates a C major seventh chord, which sounds very good, even though it wasn't intended.

The original demo had instrumental links played by a french horn; at one stage, the horn was removed and my wife preferred this version. I am considering playing the links on the vibraphone, which will add some variation without being heavy handed. We'll see.

The third song, which was described to me as a children's song, presented challenges of its own. My first act was to record the singer acapella; from this, I created a MIDI file with the tune, then added the chords. The rhythm is interesting: the chori are swung (12/8) whereas the verses are in simple time (4/4). I don't think I've ever created a track which oscillates between the two feels, so this was slightly complicated. The rhythm is carried mainly in the bass guitar, so this part varies between dotted quavers and simple quavers.

Hearing the song sung acapella brought the sound of brass bands to my mind, so I found a MIDI version of the Beatles' "Yellow submarine" and lifted one bar of the brass bar arrangement into my arrangement. Otherwise the music is arranged fairly blandly.

At the end, the singer wanted to slow the recording down, something which isn't possible to do with the version of Reason that I have. After thinking about it, I found a solution: I reduced the number of notes played in a bar, thus apparently slowing the tune down. I also had one bar of singing answered by one bar of vibes echoing the tune, thus causing a further slow down. After four bars of this, the original rhythm was reinstated for the final bar along with a 'bang!' ending (a chromatic run up from G to B). After changing key and making the track faster, the singer was very pleased with the result.

[SO: 3259; 2,11,30]

No comments: