Thursday, September 26, 2013

More holiday

I haven't been sitting idly at home for the past few days. Every day I've been studying for an hour or two; I'm bogged down in the two statistics chapters and will have to devote a great deal of time to them. Although each little part is not difficult, there are a bewildering number of different approaches and each only has one example, not explained too well. I think back to the finance course of the MBA - I had no difficulty with the material whereas most of the class was struggling. This material should be slightly easier than the finance course, which means ....

Here's an example: Kell Gardens have recorded the number of visitors over the last six weekends. The numbers are summarised in the following table

DayWeek 1Week 2Week 3Week 4Week 5Week 6

Test whether the average number visiting on a Sunday is significantly higher than on a Saturday.

As the data refer to a Saturday and Sunday each week, this has to be treated as a paired test. The factors, such as the weather, that encourage or discourage visitors are likely to be very similar on the Saturday and Sunday in the same week.

[Here is the maths part which I will spare you]

As t* lies in the centre of the distribution, it is highly likely to come from the distribution under the null hypothesis (average number visiting on a Sunday is not significantly higher than on a Saturday).

The formal decision is Reject the alternative hypothesis. The conclusion is that the average number visiting on a Sunday is not significantly higher than on a Saturday.

I suppose that I could analyse the data from my CPAP machine in the same manner to see whether the average number of apnea on Friday nights is significantly higher than on a Thursday, or whether the average number of apnea on days when I travel is significantly lower than on days when I don't travel.

I shall continue to beat my head on this material.

I have resumed work transferring cassettes of songs recorded 40 years ago. I am discovering songs which I had completely forgotten about, mainly three chord ephemera which I had probably written on the spot. One tape opened with such a piece  called "John's song"; this had me scratching my head for several reasons. First of all, I don't recall the song nor do I recall anyone called John. There's a badly recorded acoustic guitar playing chords with me singing, alongside a more cleanly recorded lead part which sounds suspiciously like me playing (I can differentiate between my playing and Robert's); could this mean that Robert was strumming while I was singing and playing lead inbetween?

The recording of the next song also had me baffled at first: where has all the hum from Robert's cassette recorder gone? Why is the song recorded in stereo (guitar on one side, vocal on the other)? Slowly it dawned on me that I must have recorded these songs on my own in London, probably in September 1974, on my newly purchased stereo tape deck with two separate microphones. The opening track was probably an attempt at multitracking, with rhythm and vocal recorded, then played back on a mono cassette recorder whilst I played along. I was to get better at this. The songs, though, were written between March and June 1973 and presumably were recorded then; what happened to those recordings?

I was monitoring the transfer from cassette to computer with headphones connected to the transfer machine, so I could hear the stereo. Unfortunately, the recording program on the computer was defined to record in mono; when I listened to the recording, I got an earful of guitar - and no vocal. I then redefined the program so I could transfer in stereo.

After chopping up the 40 minute recording into separate songs, I started cleaning up each one. At first I saved the stereo file as mono, thus combining the guitar and vocal, then resaving as stereo, so that I could have turn the recording into 'fake stereo', where a small delay is added to one track and both tracks have different equalisation settings. The result was ok, but a bit brash.

It occurred to me later on in the evening that due to the clean separation (I would record with one mike inside the guitar's sound hole, which would not pickup any vocal, and one mike for vocal, with a little guitar picked up), I could split the stereo file into two separate mono files. In my multitrack recording program, I included the same guitar track twice: once panned left, once panned right. One track has a little delay and one has some chorus, making a nice stereo sound. The vocal was panned down the middle; as this was a separate track, I could apply the same tools that I use today to improve the vocal sound: compression, equalisation and reverb. I draw the line at "tuning" the vocal!

The result is a much better sounding track than the fake stereo version, so I've been converting all the originally recorded in stereo tracks into this three track format. It takes about fifteen minutes per song, as I'm also cleaning up the beginnings and endings. These recordings are much cleaner than the Bristol ones, but they still need some editing.

I completed my Sandy Denny covers album with a recording of 'The Sea'; this is the third song which I've taken from the eponymous Fotheringay album. As I wrote earlier on this subject, Sandy's songs are always verse based (this one has four verses, one of them instrumental), but the chord sequence is unconventional and adventurous. Married to an uneven phrase structure (the first phrase is five bars long whereas the second is four; the third is four and the fourth phrase is five bars long), the song provides an interesting challenge.

As always, the first steps were the hardest: I played around for a few days with instrumentation and styles until I landed on something good. The lead guitar of the original is replaced here with a recorder, enhancing my imprint on the song. I was tempted for a while to have the recorder play the original lead guitar solo (but not the obbligato part) - this was one of the first solos that I ever worked out - but decided to go for something more original.

I estimate that I spent around 20 hours working on this song, including mixing. This is probably about the same amount of time that most recording artists devote to their songs (if not more), but it's a far cry from the amount of time that I used to devote to a recording 40 years ago. If all the songs were written, I could record 'an album' in an afternoon! Of course, there was no possibility of editing, only rerecording a song, so obviously I didn't "look at songs under a microscope", as the expression put it.

No comments: