Last Fall I was at my Grandmother’s 95th birthday party. Two of my father’s cousins, who are also sisters, had come from Ohio to join in the festivities. My grandmother, their Aunt Jean, is the last of her generation, so the cousins had lost contact, and were delighted to renew relationships with their extended family. One of the sisters is suffering from late stage cancer, but amazed us with her attitude, positivity and good humor. At some point during the festivities she talked my sister into singing a few songs for her and talked me into accompanying. We started with my sister’s favorite song, and one that seemed apropos to the situation, How Can I Keep From Singing. It went over so well that I told my sister that we HAVE to record it.
The lyrics to How Can I Keep from Singing were first published in the New York Observer in 1868 and attributed to someone named “Pauline T.” Nothing more is known about Ms. T. but Robert Wadsworth Lowry added the familiar music in 1869. Nearly 100 years later, in 1950, Doris Plenn added a few more verses that still show up in popular recordings.
When I arrived home from the birthday party (a 16-hour trip back from the Midwest where I grew up) I started listening to various recordings of the song, in an attempt to figure out what direction I wanted to take it. I heard some interesting a capella versions that were definite crowd-pleasers, but not really the direction I wanted to go. I was really moved by the Celtic Woman version, which has a minimal instrumentation and a superbly talented singer. It was too sparse for what I wanted, but served as the fundamental inspiration for my version.
In the end this is a very Celtic-influenced version of the song, with Tin Whistle and Uilleann Pipes providing the counterpoint and instrumental leads. A small Ewe percussion section provides rhythmic support through much of the piece, piano underpins the whole work, and there is also a resonant pad that changes in intensity over time. Finally, a string section comes in for the third verse providing some variety and weight to the arrangement.
The form of the arrangement is fairly straightforward. It opens with a rather grandiose instrumental introduction, followed by a very simple first verse with just piano, pad and voice, followed by a restatement of the introduction. The second verse follows adding percussion and some counterpoint with the tin whistle.
Now we get to the Chorus, followed by another instrumental section that expands harmonically on the introduction. The introduction starts on a chord a bit away from the tonic and works its way back to fundamental key. This instrumental section builds on that, starting further away from the tonic and taking a little more time to get back home.
At the third verse the string section is introduced and the piano and pad drop out. Mid-way through the Uilleann pipes provide a counter-melody. When the chorus comes in we have reached the high point of the song musically, with the string section, piano, and both winds accompanying the voice. Following the chorus is an even longer instrumental section, building on the introduction and first instrumental section.
After all this excitement we get to release, with the final verse and chorus back to just sparse piano, pad and voice. The final chorus brings the intensity up just a little bit, with the tin whistle joining in lightly, and the outro is a brief restatement of the intro. A satisfying journey, I hope.
Fortunately, my sister’s husband does some recording himself (interestingly both my brothers-in-law write and record music), and even runs an independent micro-label, which is, I think, his true musical passion. My end was a Kontakt-fest, using The Grandeur piano library, Factory Library Uilleann pipes, Factory Library Two Against the World pad, Factory Library Kroboto-Boba-Kidi percussion, Embertone Shire Whistle (Kontakt library) and Garritan Personal Orchestra strings (the version i use is Kontakt Library).
I purchased the Dicky Deegan Uilleann pipes library, but learned that it doesn’t include fingered vibrato (surprising) which is a playing style that I definitely wanted to use for this song. If I was doing a rip-roaring jig, then Dicky would be my first choice, but for this sort of legato, singing-style playing the Factory Library pipes ended up being the better option.
I made extensive use of continuous controllers for the various instruments (except the piano). The pad varies in both volume and intensity. The winds vary in attack, volume, and vibrato. Finally, the strings have volume and intensity changes. The MIDI Arrangement window to the right shows all these controller tracks.
I mixed the instrumental track down to stereo and added a queue track where I counted down to each of my sister’s entrances (since they vary throughout the piece) and sent it up to her in Vermont. This is where we hit the first snag.
My arrangement has four verses. She had mentioned to me that she only likes three of the verses, finding the fourth original verse to be too poetic to have a clearly understandable meaning, and she just didn’t care for the modern verses (which really don’t fit the context of our recording, anyway).
Hearing her concerns I took a shot at writing a fourth verse. While my results were decent, they definitely had a modern feel (the only lyric writing I do right now is contemporary worship songs, which have a definite feel and style), and my sister wasn’t enthused. However, that did inspire her to take a shot at her own verse which turned out AMAZING. With that settled her husband recorded five audio and video takes of her singing the song and sent them back to me.
Once I had the files from her recording session imported into my project I could get settled in earnest. I took the five takes my brother-in-law sent me and comped together her best performance. I took pains to use entire phrases wherever possible since I knew I would be syncing this with the provided video and wanted to avoid awkward edits. Fortunately, my BIL had the forsight to change the camera angle for each take so as I edited them together it would look a little more planned.
As is my practice with this sort of mix I started by bouncing all of the MIDI instruments down to audio tracks. I find the visual representation of the waveform easier to navigate while mixing than the piano roll bars. The waveform conveys more information I need while making mixing decisions. From the screenshot you can see that the piano and pad were rendered in stereo while everything else is in mono.
The first step in any mix after setting up the basic processing (console emulators, mild saturation and gentle compression) is to set the rough balances and panning. The piano and pad are panned slightly left and right of each other. The two solo instruments are panned further out, and the strings and panned across the stereo spectrum.
Some other brief processing notes:
- The piano has gentle shelving boosts in both the bass and treble, along with the Clariphonic for some extra shimmer that I can’t get with a standard EQ.
- The solo instruments both have CA-2A compressors and the Shire Whistle has a gentle hi shelf roll off to tame its tendency towards shrillness.
- The percussion has a hi-pass filter to help tame the bass and a gentle high shelf boost to help to instrument pop in the mix.
- The strings have all been bused to a single track so that the volume of the section as a whole can more easily be automated in the mix.
- The two reverbs I used to add space were Orilriver and Breverb 2. The Orilriver was recommended as a great free reverb plug-in, but I found (at least in Cakewalk) that it reset the settings every time I closed the program, so I had to set it back every time. It’s a shame because the ambiance was really good. Someday I have to explore Beverb beyond the Warm Orchestra patch, but I like that one so much…
The most processing was done to the most important element of the mix – the lead vocal. To be honest, my sister’s very dynamic vocal performance left me with a little bit of difficulty getting to slot just right into the mix. She brought a lot of passion to her delivery, and I wanted to save that but also make everything play nice together.
I started tackling this by using serial compression – chaining multiple compressors together, each doing something different. The first was set to a high ratio but a high threshold, catching the highest peaks and bringing them back to earth. A follow-up compressor was set to a lower ratio and threshold, reigning in the entire performance a bit. But I just wasn’t happy with the results. It was either too controlled and lifeless, or still too dynamic and popping out everywhere. I couldn’t find the balance.
Then I remembered an older processor that I haven’t used in a few years, the VOS Slick HDR. Described as a Psychoacoustic Dynamic Processor, it is essentially three chained compressors that feed back into each other. As seen in the screenshot, this is what made it into the mix, and did the job perfectly. Other processing on the track included a high-pass filter to remove any extraneous rumble or hum from the recording, and the Clariphonic working its high-end magic. Once all that had been applied the track was starting to get a little sibilant, so I tried several de-essers to clean it up. The one that ended up working best was the Auburn Sounds Couture. While it is technically a transient shaper, the secret is that it also works really good as a de-esser.
The master buss processing should look really familiar if you’ve read and of my earlier posts. In order:
- Tape Emulator
- Console Emulator
- Quad-Curve EQ (with a gentle low roll-off because the mix was still a bit muddy)
- Density MK III compressor in mid-side mode
- Blue Cat Gain (to set the final level precisely)
- Unlimited Limiter
- MLoudness Analyzer by Melda Productions
I set levels for YouTube, so I’m shooting for -14 LUFS. Some would argue that you shouldn’t aim for specific LUFS targets, but that you should just make your music sound the best you can and “right” for you and your vision. I’m fine with the concept, but I could make this style of music really super-dynamic. By using the the -14 LUFS target with peaks no higher than -1, I can set the dynamic range knowing that my listeners (both of them) will find the experience in the sweet spot. YMMV.
As mentioned above, the ultimate end for all the music I work on these days is a YouTube video. Interestingly, YouTube is currently the world’s largest music discovery site. Anyway, my concept for the video featured the footage my BIL shot of my sister singing her parts. For the lengthy instrumental interludes I inserted slideshows. The first instrumental has photos of the instruments featured in the arrangement; the second has shots of thundering storms; the third is raging seas; and the fourth and longest is scenes of faith, such as churches, bibles and other such themes. At the end the instrumental outro just has a photo of Karen, to whom the song is dedicated.
The first step was to take the original footage sent to me by my BIL and import the clips into DaVinci Resolve. There I ran them through color correction and color grading. The original footage sent to me had a very cool cast, likely due to the lighting used during the filming. I consider myself lucky that he could film and send it to me – I’m not going to fuss about lighting.
Then, since I wanted to feature the slideshow, I brought everything into the discontinued Windows Movie Maker. If you search you can still find it and I haven’t yet found anything easier for creating slideshow movies. It’s a bit harder to sync the audio and video, but not so much as to be unworkable. I really need to set up some presets in DaVinci Resolve to mimic the effect I can do so effortlessly in WMM – maybe next time.
When the movie was complete in WMM, I rendered it and brought it back into Resolve. I did this for two reasons – first, I wanted the audio for the All Things New Productions tag at the end to seamlessly blend with the fadeout of the main video, which I don’t think you can do in WMM. You can make video clips overlap and fade, but pure audio clips just butt up against each other and I cannot figure out how to make them fade.
The second reason is that I want to make sure the video renders with 24 bit, 48k audio quality, and I cannot guarantee that from WMM. Once everything was lined up and balanced I rendered the video and sent it up to YouTube, and that, of course, is what is up at the top of this post. I hope you enjoy – the next Laus Deo release will be a medley of gorgeous minor-key Christmas carols. However, there are a couple of Community Worship Collective songs in the queue first, so hopefully later this summer or in the fall.