On Thursday, April 16th, 2020, Lucy shared an ITV news segment of a moving tribute to our wonderful National Health Service (NHS) workers and those who have lost their lives to COVID-19 here in the UK. Apart from making her tear up just a tad, the main reason she shared it was because it was something she worked on with her husband Phil, the Musical Director of The Lewisham & Greenwich NHS Choir (the one waving his arms about in the video). In this article, Lucy explains how it came about and how they managed to pull it off. Over to Lucy.
The track throughout the video was described as the choir “performing a special rendition of their number one single”. Although not technically incorrect, I'm not sure it came across clearly that this was recorded and mixed in the two days prior, with everybody in isolation. But anyway, Mike asked me to write about the process that Phil and I went through, and I will do my best to explain what I did, how, and why.
A New Skill And A Quick Turnaround!
Although I mixed a song the other week for the NHS choir, as well as a few tracks for my own choir The Adam Street Singers, I would certainly not classify myself as a Music Producer/Mixer by any means, not professionally anyway. What I have done is to combine my post-production engineering knowledge with my musicality (thank you classical music degree) and have done my best in the small time frame we had. So this is definitely not a how-to guide on mixing choirs!
As with most TV work, there was a really quick turnaround for this – we started working on it Tuesday morning (from the re-arranging stage) and delivered it on Wednesday lunchtime. This meant that I didn't have all the time in the world to edit and mix this, but also the choir didn't have very long to learn it (if they didn't already know it) and record their takes to then send to me.
The Background
I'll start at the beginning. The NHS Choir were approached by ITN (Independent Television News) to record something for their NHS tribute broadcast, presumably, they knew it was something that could be done because a week previously they had been a news report about a virtual wedding for two of the choir members and 'performed' a song over Zoom that they would have sung for them at their actual real-life wedding. I put 'performed' in inverted commas because, as anyone who has tried to collaborate musically over Zoom will know, the latency of video calls over the internet creates an issue, with everyone coming through your computer/phone out of time with each other, despite singing or playing IN TIME their end. So although the choir was all singing along on the Zoom wedding, they were muted and the voices heard were pre-recorded (and mixed by me). They were, however, still singing from home in both the Zoom video and the audio track.
Get The Parts Out
Fast forward to this week and my husband Phil was asked to cut down their 5-minute song “A Bridge Over You” to 3 minutes, which meant he had to then teach the choir what they needed to sing as it would be slightly different from what they knew. Add to that the complication that the newer choir members have never sung this piece so he had to teach them their parts quickly. In addition, not all of the choir reads sheet music so he couldn't just send an adjusted PDF out for them to learn. So we did what we have been doing for their 'Virtual Rehearsals' over the last few weeks and sent them “The Choir of Lucy and Phil”. This is, as I'm sure you have already guessed, a mini mix of me singing all 4 female parts, and Phil singing all 4 male parts.
In order for the choir to learn their parts, and then record themselves we needed a track laid down that would be quick and easy for them to pick up what they should be singing and record something for us just as quickly and easily. For the final choir mix track, we would be mixing in a lovely piano track performed by their Accompanist/Assistant Musical Director Liam Dunachie, but as I mentioned, we didn't have very long to do this all and Phil needed to get the choir learning and recording as soon as possible so we weren't up against it at the Wednesday lunchtime deadline.
So he sent the amended sheet music to Liam, along with a Logic file containing a click track with all the tempo changes already in place and a few MIDI piano notes indicating a few of the vocal lines for guidance (made by hand on the computer as he had lent his MIDI cable to someone pre-lockdown!).
Prep The Tracks
In the meantime, Phil added a simplified piano track for the choir to sing along to, so we could get them practising and they could sing more accurately with less going on in the piano. (Unimportantly, because Phil didn't have his MIDI cable, we had to connect his keyboard to my audio interface for the first time and record audio straight into Pro Tools which we'd never done before – luckily I had a jack to XLR cable in the cupboard! Not sure why I did but hey ho!)
Phil and I recorded our guide vocal parts, being really accurate with rhythms, phrase entrances and endings, specific breathing and general musicality like dynamics, pronunciation, and detaching certain things and making others more legato (thank goodness for my sight-reading skills!). We recorded a few takes of each part so it sounded more like a group of people singing and I even threw in a bit of reverb and a basic doubler on an Aux Track with just the vocals routed to it, to help with this (mainly to hide my not-so-great soprano singing...!)
I bounced out each individual vocal part with the simplified piano part and click track so that they didn't have all the other parts confusing things, could hear all their ins and outs/breathing etc, and also so they could, if they needed to, learn their individual parts 'Parrot Fashion' accurately and in time with the click. This was sent along with a PDF of the amended sheet music.
The Record Session!
We asked them to play their tracks on one device, listening with headphones, and record themselves singing on another. Because the end section splits into '2 choirs' (where the two songs are mashed up) he asked them to send two full takes of the song, one with with 'Choir 1's part at the end, one with 'Choir 2's (rather than just sending a separate smaller clip of just the end section) so we had more voices to bulk out the track, in case we didn't get many of the choir sending clips through. (They are of course all NHS workers, so the majority of them are working really long hours just now! Obviously this is useful for having a richer and fuller choir sound, but of course, gave me double the work!
Those who knew the original arrangement well sent their recordings through on Tuesday afternoon, but the choir had a rehearsal Tuesday evening (filmed for the ITV Video), and as expected, a lot more recordings came through Tuesday night and Wednesday morning. So that I wasn't constantly going over the same bits and rebalancing the voices within each part, I worked on it in two waves; one on Tuesday afternoon/evening and the second on Wednesday morning.
The Mix - Day 1
Phil very kindly split the recordings into folders for each vocal part for me before handing them over, so I could at least import parts one at a time and work on them as a whole before importing the next vocal part. To give you an idea of scale, in total I had 33 choir members' recordings (each with 2 takes for Choir 1 and 2) and then 2 takes on each part from either me or Phil (in the mashup section, 4 from us on each vocal part!). So that made 66 phone/laptop tracks, with 8 (and sometimes 16) studio-recorded tracks from me and Phil. So mostly 74 audio tracks to sync and mix.
Import Audio
He got sent all sorts of files – mp3, wav, mp4, m4a, even Opus files (had to google those!). Some stereo, others mono, some quiet, some really loud. They also came in varying sample rates – I learned this the hard way... Out of habit, I tend to just drag files from Finder or Soundminer directly into Pro Tools to import anything. However, when I did this with some of these files, they were coming in at different speeds and keys... I was confused at first but thought about WHY this would happen. SAMPLE RATES!
I then went to import them to Pro Tools via the Import Audio window and saw that, as expected, some of these tracks actually had a sample rate of 44.1KHz (mainly the mp3s) and my session was 48KHz. I had assumed that Pro Tools converted sample rates to target the session's format when you drag and dropped into Pro Tools, but turns out not! (at least not with my settings, perhaps I can change this, will look into it). Anyway, I converted them all within the Import Audio window and it solved the issue. I did have to convert the Opus file on an online file converter website however as Pro Tools didn't recognize this as an audio file.
Sync Up
Before I did any sort of mixing I lined all the takes up and made sure it was all in sync. If we were to do this again Phil and I have said we'd introduce a sort of 'clapper board' system – where the choir clap on one of the beats of the bar of the click before the singing begins. Then it would be easier and quicker to sync up. Instead, I lined up everyone's opening line with mine or Phil's depending on their part, but sometimes I found this up meant later bits were slightly out of sync. This is absolutely not a criticism of the choir – I've realized that people vocalize their consonants at different lengths at the beginning of worlds, completely unintentionally – so although the main vowel sound can be correct, the consonant can be slightly earlier or later than other people's and it's not something Phil would have had time to perfect with the choir.
Once I synced takes for one part, let's say Sopranos in this case, I then skipped through to each section of the music to check the rest was in sync. As a whole the vocal lines were in sync/in the right place, it was mostly ins and outs the needed fine-tuning (so to speak – not actual TUNING, just timing-wise), particularly on ends of phrases with words with S or hard consonants like D. There's also the inevitability in his type of song of rushing in some areas. They didn't have Phil conducting as they would normally, and he couldn't fix it in real-time if any such rushes occurred, so there were a few bits that I needed to neaten up with editing. Considering they all sent their recordings in one long take they did really well. They didn't have long to learn their parts, Phil couldn't give feedback on the intricacies of the musicality, and it's hard to get every single thing super accurate in one take, especially when you are busy saving lives and trying to spend time with your family! So again, this was to be expected and absolutely not a criticism, I was really impressed with what we were sent.
Clean-Up Part 1
I deleted any long periods of 'silence' to remove any erroneous movement, clicks, coughs, page turns, breaths, etc, and made sure to fade in and out on any edits. This sounds obvious, but with higher quality recordings and small auto-fade settings on Pro Tools, you can sometimes get away with hard cuts. But with this many phone and laptop recordings, even those that sounded 'cleaner', the 'silence' is much louder than you're used to because of the slightly raised noise floor, multiplied by 66! I did delete some breaths for the same reason – either removing all breaths completely, like the opening line, or just leaving a few in and clip gaining them down. Leaving breaths in rather than removing them keeps the natural performance, but 74 breaths all at once are incredibly loud and intrusive!
Before moving onto the next vocal part, I balanced the individual soprano voices against each other using Clip Gain for speed (I did not have time to ride faders on absolutely every track!).
Noise Reduction 1
At this point, I did any noise reduction required on some particularly 'noisy' recordings using RX7 Advanced (mostly Dialogue Isolate and Voice De-noise) via RX Connect and rendered this to the file itself. One file oddly came in really distorted and compressed for some reason, so I did a few stages of 'de-clip' on top – didn't fix it entirely, but certainly helped.
On some recordings I needed to do some specific EQ-ing, mostly because they sounded too muffled; perhaps this was the closeness to the recording device, or purely just the quality of their device's microphone. For this, I just used Avid's EQ7 as an insert on these individual tracks. I didn't push up the high mids to counteract this as the noise floor on these recordings was higher than your usual sound studio recording, and a lot of that noise lives in those frequencies so boosting them would make them hissier and require more de-noising, which seems counterproductive. So I mostly just reduced the low mids, around 300Hz and that helped a lot. Some had low wind or rumbly rooms so I gave these tracks basic high-pass filters, and in general, lowered the bass shelving EQ. I didn't EQ and de-noise every single track to make everything as clean as possible as I knew I wouldn't have time, and didn't know how many more recordings I'd be receiving, so I only individually treated the ones with more extreme issues.
A few files came in incredibly bright, with more high hiss also, perhaps these ones were recorded through a laptop instead of a phone? On these, I brought down the high mids and also added low pass filtering, occasionally boosting their chestiness in the low mids, but only a little! Again, I didn't go to town on individual tracks, I only fixed the big issues that stood out, as I knew I would have a lot of tracks, and therefore the individual tracks would be played at a lower level if needed. You can see how many/few tracks I did this to in my screenshots of the final session – on 'INSERT A'.
Colour Coding
I repeated all of this for all vocal parts, and color-coordinated everything so I could easily adjust certain vocal parts. I grouped all tracks in one vocal part together using Command+G and balanced the vocal parts as a whole against each other using Clip Gain. If I were to do this in the future I would set up Aux tracks or VCAs for this, but I was trying to be quick, didn't have a template ready to go, and didn't want to spend time setting up busses and sorting out lots of routing. Plus, with my editing experience, I am pretty quick with a mouse and keyboard!
The Proper Piano Part
Liam had sent his piano track by this point so I imported that and muted the click and skeleton piano tracks. I left the click in place so I could see visually where certain phrases should land.
I kept the doubler and reverb I had from the guide track, (Waves Doubler and Avid's ReverbOne) and just tweaked the settings, mostly just the wet/dry level of the reverb. I didn't have a lot of time to find the perfect reverb, but with so many voices, a full piano track, and reverb wetness of around 21%, the difference wouldn't be heard too much. I was more concerned with making sure everyone was in time with each other and the balance of the vocal parts was optimal. Once routed to this 'Choir' aux, I vaguely balanced the vocals against the piano, not spending too long perfecting it as I knew I would get more voices the next day and have to remix anyway!
I loved Liam's new piano sound but as I worked with the track I just wanted it sound a little more full and rich so inserted an eq achieve this.
I then realised I wanted a bit of reverb and sent this track to a Piano Aux via Bus 3-4 where I inserted ReverbOne with just 10% wet. I usually use sends for reverbs so I have more control, but knew I'd just be setting them for the voice and piano and then leaving them the same throughout. I could have made this neater by inserting the reverb and EQ on the same tracks for ease of navigation but was just adding things as I went along and didn't want to spend too much time re-arranging thing and tidying things up.
1st Review - Tuesday Evening
Before I turned in for the night on Tuesday evening, I listened through to what I had so far, with vague volume levels, basic individual EQs and noise reduction, together with my global doubler and reverb on the vocals aux. It wasn't bad, but there were certain things that were sticking out for me. Because of the quality of the microphones (namely phones and laptops), the brilliance of their voices (ie above 5/6KHz) seemed to be really prevalent when all played together and everything suddenly sounded more hissy/noisy.
More EQ, De-essers And Gullfoss
The sibilance hurt my ears and again the breaths were intruding a lot in these frequencies. I inserted 2 different De-essers onto the Choir Aux track and added my trusty EQ7 with a low pass filter and lower high-frequency shelf. I also added to this EQ a high pass filter to eradicate the bassy plosives, mainly on the word “Bridge” which is sung quite a few times. No pop shield + low-quality microphones = some really windy Bs and Ps!
I then added an auto EQ plug-in on my master bus called Gullfoss, with vaguely set settings. I haven't really played around with this EQ yet, so I probably am not getting the most out of it, but I just wanted to chuck something on to see if it can make things sound nice overall with not much extra work because of my deadline.
Obviously the increased high end from this contradicts the low-pass filters on my vocals – but the master fader has everything including the piano and reverbs and I didn't want to spend too long fiddling about with it.
Day 2 - Wednesday Morning
The next morning I did the same things all over again with all the new recordings and then listened to the whole thing in full, with all vocals parts and the final piano track. Because of the issues, I mentioned earlier of an increased noise floor with 66 tracks of low res recordings, I decided to create an Aux Track for noise reduction and routed all the vocals to this first (via Bus 1-2), which then was routed to my 'Choir' Aux Track that has the EQs and reverbs (via Bus 5-6).
Noise Reduction 2
I used iZotope’s Voice De-noise as an insert here, with a noise reduction level of just 4dB, being careful not to overdo this as some tracks weren't that bad in isolation, and moved my de-essers onto this Aux Track just to keep everything in the same place. I also noticed that a lot of the 'D's at the ends of phrases were sounding a bit like 'T's, and some lip smacks/general clickiness were coming through so I inserted RX7 Mouth De-click. I don't usually just use BUS numbers, I normally rename my Busses in the I/O setup window and have my sessions lovely and organized, but I was concerned about time!
Aux Tracks Processing
You will notice on my Choir Aux I have a send going to Bus 7-8. This is to another Reverb Aux Track, which I put in place right at the end of my mix to soften the ending. The reverbs on the vocals and piano were very light, just to give a tiny sense of space, so didn't give the effect I needed at this point. I set this extra reverb to 100% Wet, with the track's volume at 0dB, and simply only sent the choir and piano to this track right at the end, automating the send in a gradual incline rather than just mute automation which creates a sudden jump.
Below shows my 'send' automation on these tracks.
To give or more succinct idea of my routing, I have taken a screenshot of my mix window for the aux tracks.
Conclusion
So there you have it. My methods were not perfect, and should I do this again, I will hopefully either have a template built or have the time to route things in a more effective and efficient way. As most sound engineers do, I was criticising my mix when watching it on telly or online “I should have mixed the basses lower there”, “you can't hear the altos very clearly in that bit”, “those T's are a bit in your face here” etc... and I thought I had mixed voices in the “Sail on Silver Girl” section louder than they came across; maybe I didn't or maybe the transmission compression or the online uploads did something to my mix. I didn't even think about loudness when I delivered this, with my music head on, completely forgetting that it was going on TV and just thinking “this sounds nice”, so my dynamic range may have been a bit too much for broadcast! But I am pleased with what we managed to achieve in such a short space of time, with less than ideal recording environments and equipment.
The Final Product - NHS Choir's Tribute To Those Who Have Given Their Lives To Save Others
In tribute to all those who have given their lives to save others, the Lewisham and Greenwich NHS Choir have sung a special rendition of their number one hit; "A Bridge Over You" for ITV News.
Official Government figures say, 27 nurses, doctors and care staff have died from coronavirus, but announcements from NHS trusts and tributes from loved ones indicate the true number is higher still, with more than 40 NHS staff thought to have died with COVID-19.
The tribute, which featured on Thursday's ITV Evening News, comes as the channel dedicated the day to NHS staff fighting COVID-19.
And for the fourth week in a row, millions of people across the United Kingdom are expected to show their appreciation for NHS and key workers with a mass round of applause from doorsteps, balconies and open windows.