The wedding card song is such a key aspect of the installation. In order to do its job within the whole, it needs to be melodic yet unnerving, apparently unfinished yet professional-sounding, somewhere between a jingle and a nightmare. The first recording (here) was made using a melody composed from remembered snippets of nursery rhymes and hymns, which I recorded as a single track and then played into my headphones to improvise some harmonies. I made around 12 tracks of possible harmonies, and chose two of these to layer over the original melody, using a process of trialling combinations.
Having listened to that original three-layer sketch upwards of 30 times now, I know two vital things, the first of which is that my singing voice is weak. While that may be a quality that could be exploited for other purposes, this recording needs to sound professional if it is to create the illusion of a remembered fragment that keeps returning, like an earworm. So, on the 24th I have five capable female singers meeting me to record a version. I may not need as many layers as that, but it will be good to have the choice. Secondly, the melody resolves too neatly at the end. The feeling of the song should be that it is part of something larger which has been fragmented, which does not happen when it ties up like this. So, I've spent the morning with Tom listening to him talk about why the song currently sounds like it resolves, watching him instantly find all the parts that I improvised firstly on his guitar, then translated onto a keyboard so that he could play two at a time, and working out together how to make the song end in an open way, so that the listener feels it hasn't quite completed without actually cutting off mid-phrase. As a result, I now have several versions of each vocal part to trial on the 24th. These are currently in the form of rough recordings of the keyboard, which I will translate into voice in order to communicate to the singers, as I'm not able to write musical notation. When the recordings are made, and the files layered up to create the song, I'll take the finished audio file to the engineer who is going to show me how to use the audio to lathe cut a vinyl record. The machine we'll be using is a lathe that 'hears' audio and translates it into a cut, so it is vital that the audio file is right before we start. Unlike commercial processes, where a metal template is made which many records can be pressed from, this machine will make a unique disc. Incidentally, in my talks with the people involved in getting access to the lathe, I've learnt that 'ghosting' is a term used for when the lathe 'hears' something unintended and cuts an extra sound into the vinyl. I'm sort of hoping it will happen.