This one was more difficult than expected.
How Great Thou Art started as a Swedish poem in the late 1800’s, written by Carl Gustav Boberg. The melody is a Swedish folk tune, and the song grew to popularity as it was used for the Billy Graham Crusades.
My version started as an idea – I wanted to take a hymn that is often associated with booming voices and thundering organs, and present it in a simple, contemplative arrangement. I toyed with the idea of string quartet, but eventually settled on solo violin with piano accompaniment.
My good friend Dena tracked the violin part last summer when she was up visiting. We ran into a bit of a technical glitch. As mentioned in the earlier post, I didn’t have an adequate keyboard to accompany her. As a result, I took one of her two takes and practiced playing along with it until my piano recording session last Fall. I ended up with two takes all the way through and several attempts at the ending (of which I ended up keeping the very first take). Actually, I used the first take for most of the final comp, except for a few places where the second take had slightly better rhythm.
The violin track, at the top of the screenshot, has three different colors. The light blue is her first version, which I played along with. The dark blue is the second version, which was at a slightly different tempo, making editing difficult. The yellow/orange bits are sections where I went in and digitally adjusted the timing. (The piano is all one color for two reasons: first, it was all one long recording, and I didn’t change the color when I split up the takes; and second, some of the visible breaks are within the same take – see below).
Comping this all together proved to be a challenging project. First, when Dena recorded the original violin line, she was playing against a very poorly played keyboard. Being the good live musician that she is, she did her best to compensate to make the whole as good as possible. The unfortunate side effect of that is that her take, all by itself, was not entirely smooth. In retrospect I wish I had gone in an cleaned the part up a bit before recording the piano to it. But I didn’t.
In the end, I trimmed both the piano and violin in numerous places and pushed and pulled them around until we were playing both smoothly and together. Many times the violin had slowed or held a note waiting for the keyboard to catch up. Consequently, the piano then had to wait for the violin. All that was taken up in the editing. Where actual cutting was not practical, I split out short clips of the violin (because it was recorded in mono, rather than stereo) and used V-Vocal to skootch the timing ahead or behind until everything was tight (or at least not sloppy).
Lastly, the places where the second take was cut into the first are times when there was some sort of an audio glitch, and I needed to pull from a good take just as a short replacement.
To the left is a screenshot of the processing applied to the violin recording. First, the lows are rolled off using the channel EQ. Then into Slate VCC for channel emulation. From there I inserted vladg’s Proximity to try and pull some of the room out of the recording. I’m just not that happy with the recording of the violin and might even go so far as to try and re-record it this summer if Miss Gaddie comes visiting again. The last two in the chain are Bootsie’s Thrillseeker LA compressor, taking just a little off the top and adding just a touch of saturation, and his BootEQII, boosting the highs and dipping a little of the low mids for some tone shaping, with another thin layer of saturation.
There is a touch of automation on the track as well. The green envelope up in the tracking window is the send to the reverb, and if you look closely, you can see a piece of clip automation where I manipulated the volume of a Take 2 clip to make it blend better with the Take 1 clips on either side of it.
The piano receives similar processing to the violin (but without Proximity or channel EQ), but with different plugins. After the RC-Tube channel, I inserted the Antress VP-ME, which is their Thermionic Culture Phoenix clone. I honestly have no idea if the various Antress clones work anything like the hardware they are supposed to be emulating, but this one is working well enough for me that I have been using it quite a bit for piano lately. (The Antress bundle seems to change regularly, and the VP-ME appears to no longer be a part of it. If you are interested in a copy, drop me a line – it is freeware, after all).
Last in the chain is the Sonimus SonEQ. I just wanted very simple tone shaping and a little saturation, and this plug delivers exactly that. It also found rave reviews on several on-line forums. All I can say is that it worked nicely for me.
Choosing a long-tail reverb is quite a challenge. Regular readers know that I’ve been a recent fan of the ValhallaRoom reverb, and found it quite useful for a number of spaces. However, I’ve been less impressed with the really long-tail reverbs. It’s got just of a bit of a metallic ringing when the tail is exposed that I couldn’t live with.
As a result, I went back to my trusty old Impulse Response reverb, SIR, and started rummaging through impulse responses. the biggest difficulty with using a convolution reverb is that you are limited to the specific IR’s you have in your library, and I have a lot – including a lot of pretty useless ones. To give an idea, I just scanned my IR folder and it is reporting nearly 3000 files. I suppose I should go through a weed a lot of them out, but it’s hard to know what will or won’t work in any specific situation.
In the end, I went with an IR from Echochamber – the timeworkschurchdark. The website is German, but the Google translation will get you close enough to find what you’re looking for. This IR is a little over 5 seconds long, and all I did was roll off a little of the low end so that it wouldn’t build up and get muddy. Note also that since the reverb is on its own auxiliary bus, I’m not sending any dry signal through – just fully effected.
If you’ve made it this far, you may be curious how I processed the stereo outputs. If you’ve read any of my other posts about mixing, this will look (probably) disappointingly familiar. The Slate RC-Tube mixbus for channel saturation, into Density MkIII for light compression and sonic glue, to the Boost11 limiter to bring the level up. In this case, I’m not actually limiting the signal at all, just using the limiter to boost the level. Finally, I always have my VU meters, the PSP VintageMeter, helping me set levels appropriately.
I learned a lot about editing in this mix. I also learned the importance of proper preparation when people are playing together – and about the importance of thinking outside of the box. If I could do things differently, I wouldn’t have felt compelled to record in my house. I would have looked for an off-site location (probably the church where I recorded cello last week) and recorded with a little more leisure. In the end I am pleased with the results, but not ecstatic.
The good news is that with publishing this song, I have only THREE tracks left for the whole project. My Jesus, I Love Thee is now completely recorded (as of last Monday) and just needs to be edited and mixed. Oh, the Deep, Deep Love of Jesus is mostly arranged and I’m working on setting up recording sessions with the hand-bell choir. All that’s left to write is the solo instrument and that will be recorded last anyway. Finally, The Peace Medley is well on the way, but I want to revisit the acoustic guitar part to make it more like how a guitar would play it rather than how a keyboard would play it. Then I can send it off to my good friend who will review it for me.
And somewhere in all that I have to set up three websites and a web store and do the artwork for the CD. Fun, fun!