Having just finished week one of a two week production of Shout! (last week in The Paget Rooms, Penarth, this week in the Borough Theatre, Abergavenny), I wanted to talk a little about some of my background workings as a Musical Director. While during show week my job comprises of running vocal warmups, leading the band from the keyboard and trying to cue things correctly, there’s a lot of work that gets us to that point. In a show as vocally rich as Shout!, a production where we’re using nine singers to cover five lines of vocal parts, I wanted to share how I went about making personalised rehearsal recordings for each member of the cast, using Google Sheets, Dorico, Cubase, DaVinci Resolve and YouTube.
To make guide tracks for each of the nine singers, for each song they’re are singing. This will result in nine “singer + piano” tracks, one track of just the piano part, and one track with all the parts together. These have to be easily accessible and usable by the singers.
One thing to highlight is that while I was working on this, I’d intended to use my own website to host the guide tracks, but I was having issues with the playback bar (now fixed!). I was also considering using SoundCloud, but other clients have run into issues with it, and at the time SoundCloud themselves were running into financial difficulties. To that end I decided that YouTube would be the hosting platform of choice; while they don’t allow you to host audio files, it’s fairly easy to make a video from an audio file by using a still image (leaving you with one static image while the audio plays). This would also allow me to make playlists so each individual could focus on the files that they needed.
The workflow for making each song’s guide tracks went like this:
- Designate parts (Google Sheets)
- Type up score into MIDI (Dorico)
- Make audio recordings of each part (Cubase)
- Make videos of the recordings (DaVinci Resolve)
- Upload the videos into personalised playlists (YouTube)
1) Designate Parts (Google Sheets)
The first step was to work out who’d sing what. With five named parts (Red, Blue, Yellow, Green, Orange) and four members of the chorus, we decided to set the ‘high-low’ distribution of vocal parts (from 1 = high soprano to 9 = low alto) based on the ranges of the women cast in named roles, with a chorus member in between. In our case the parts worked out as:
Red (1), Chorus (2), Blue (3), Chorus (4), Yellow (5), Chorus (6), Green (7), Chorus (8), Orange(9).
From then on it was just a case of who was singing what solo at what bar (as set by the score), and then assigning the harmonies accordingly; if Yellow (5) was singing a solo and had a two part harmony chorus supporting her, then singers 1-4 would sing the top line and singers 6-9 would sing the bottom. With a soloist singing over three part harmony we’d typically have soloist, 3 voices on the top, 3 voices on middle and 2 on bottom. This then lead to spreadsheets like the below being made:
2) Type up score into MIDI (Dorico)
Now that the vocal parts had been organised (i.e. who was singing what) it was a case of taking the notes from the page and turning them into something that could be played back by a computer. In this instance I used the notation program Dorico to do this.
For flexibility of arrangement the score was typed up into ‘Solo’, ‘Harmony 1-4’ and ‘Piano’ lines, before eventually copy/pasting the various solo/harmony lines into their specific voices (so if Red had the solo then I’d copy/paste the solo line into her part). This gave me a great understanding for what each individual person was singing, and, as I knew each of the singers, allowed me to make sure that each part best complimented the singer who would have to eventually learn and perform it.
N.B. Having recently cross-graded to Steinberg’s Dorico I was delighted to finally have a large project to use it with, and I’ll definitely be writing a more indepth account of my own learning curve with it.
Dorico screenshot - nine individual singer staves, one solo stave, four harmony staves and a piano part
3) Make audio recordings of each part (Cubase)
After exporting the MIDI files from Dorico into Cubase I could then make separate audio files of each track. As there were so many songs to be made, I ended up creating a template which I could import the MIDI into, so all the tracks would have the same ending sound. I also ended up using a compressor (iZotope’s Nectar 2) to try to help balance the changing quantities of voices at different times (specifically for the ‘Full’ recording with all nine digital singers with piano).
The end goal was to create 11 audio files (9 individual singers, the piano on its own, and everything all together) to eventually be used by the singers to rehearse with. Luckily, Cubase Pro 9 has some great features to help with this.
The process for making the specific audio files from all the MIDI went like this:
- Select the entire song (setting the locators around the beginning and end of the MIDI)
- Make 11 copies of the song; each one of these will be used for a different arrangement of the track (i.e. 1-9: each of the individual singers, 10: piano only, 11: full ensemble)
- Make a Cycle Marker around each of the copies of the song and then name the Cycle Markers based on Mute any regions not needed in each copy of the song (i.e. for “1. Red” - sung by Cat - all other vocal parts were muted, leaving only her part and the piano)
- Export all the audio files based on their Cycle Marker (using File>Export>Audio Mixdown, Export Cycle Regions)
Using ‘Cycle Markers’ on the Marker Track with ‘Export Cycle Markers’ option in Cubase greatly helped to automate the entire process; previously I’d have to select the range, solo the instruments I wanted to hear, then render the audio, but this option allowed me to set it all up beforehand, click ‘go’ and make a cup of tea while it was working.
This left me with 11 audio files, named based on their Cycle Marker, ready to be turned into a video.
Cubase screenshot - MIDI exported from Dorico goes into a digital template matching the Shout! ensemble
4) Make videos of the recordings (DaVinci Resolve)
For reasons mentioned above I decided to use YouTube as the hosting platform for the backing tracks, which meant converting the audio files into videos. DaVinci Resolve is a great, free piece of video editing software which made the entire process of moving from audio to video very painless; it even has a render preset for YouTube itself, so those with as little video understanding as myself are still able to use it.
Within DaVinci it was a case of making 11 individual timelines, then naming and rendering them accordingly.
DaVinci Resolve - rendering the audio from Cubase onto a still image ready to be uploaded to YouTube
5) Upload the videos into personalised playlists (YouTube)
The final step was to upload the videos themselves to the cloud where any cast member could access them (though not for public use). YouTube makes uploading multiple videos very easy; just a case of click and drag. I decided to make 11 separate playlists (1-9 for each singer, 10 for the just the piano parts, 11 for all parts) and then assigned each newly uploaded video to each playlist. The videos themselves were uploaded as Unlisted videos to ensure no public access, and the cast members were given a list of URLs to access each of the playlists at their leisure.
End result: For each song which involved full cast singing, 11 separate videos were made showing each of the parts.
Having heard one week of the results, the hard work was definitely worth it.