Getting to grips with Zoom

Last week Steve wrote about the video made by Cottontown Chorus as a tribute to NHS and key workers. It raised over £1700, on the day, for Comic Relief and Children In Need.

He wrote about it from his perspective. We have since been asked for more information about how the video was put together. A few members of the chorus have written about how it went for them. Some of these accounts are very long but worth reading if you are interested in doing a similar thing. They highlight some of the pitfalls as well as the amount of camaraderie.

James McCollom wrote:
Back when everyone was worried about the lack of toilet paper, I’d just been sent home to work. Could I find a webcam in any of the electronic stores? Not on my nelly. So my first few Zoom rehearsals had been via my Android phone, which has been suffering due to the constant battery use. I finally downloaded an app to connect the phone on your camera to your PC via wifi, which allowed me to see more than four people in a Zoom room at any one time! However, during the shooting of the video, we nearly had to scrub take two when my until-now-not-very-tech-savvy parents tried to start a video conference. A lucky escape: remember to learn the difference between airplane mode and do not disturb!

I’ve been hanging a (lime) green bedsheet behind me so that the virtual beach background would work properly. It’s still hanging limp from the curtain rail in my back bedroom, ready to be deployed in a green screen emergency.

I recorded two full takes after our Tuesday evening rehearsal, but must have started another twenty which I abandoned mid way through for various reasons. Another one the next morning – but in the end, the first take was the best.

My recording posture for this project was to listen to the backing track from my PC through the left earpiece of a phone hands-free kit, with my mobile laid on a shelf in front of me at chest height, singing about 15 inches from its inbuilt receiver. The sound quality was ok. Don’t be afraid to improvise. Don’t let perfection be the enemy of the good.

Thankfully no interruptions from the neighbours – they have been extremely patient with the amount of singing coming through their wall these past few weeks!

Alex M sent photos of his ‘recording studio’. His comment was “I thought my setup to get the beach backdrop working (I needed a solid colour behind me) might be amusing.”

He obviously had to sit on the floor for the filming (though I am sure he would not have been sitting down when he made the voice recording).

Christopher Price wrote:
The music
I was already confident of the notes for my line, so no extra learning was needed… until I sang along with the track. There are passages where the rhythms I’ve been singing all my life are wrong, and this only comes into focus when you’re having to copy the recording exactly. I’ve even gone back to the Beach Boys’ original and yes, the same rhythms are on there. Just goes to show, you don’t know a song until you really listen to it. I further realised on listening back to my recording that I was a fraction too late on some of my entries – not a rhythm issue, more down to the breathing and anticipation of the start of the line. I re-recorded a couple of times to improve it, but my final submission didn’t eliminate it completely. Going through this recording process will make us better individually, and a much tighter unit when we get back together in person.

The beach
We were asked to find a beach background; I found the perfect one, a single palm tree off to one side, with an azure blue sky and sandy beach. I tested it in zoom and it looked wonderful. Then we were asked to use a copyright free photo. A reverse image search revealed it was very much copyrighted. I spent an inordinate amount of time on a couple of royalty free image sites looking for a picture as close as possible to the one I was previously so happy with. It’s the little details that I think make a video like this a success and it’s up to us as individuals to take care of them.

The outfit
The instruction was for a Hawaiian shirt in line with the clean cut, laid back feel of the Beach Boys. My dressing up box let me down on this one and without the ability to borrow one, I had to improvise. My getup consisted of a striking pink beach towel, sunglasses and a hat. I hoped no one would object to me showing too much chest.

The tech
I’m fortunate to have a decent laptop, this meant that when I put the beach background on, it was pretty painless. My computer can do the virtual greenscreen. Others not so well. The chorus ran a test room prior to the recording so you could test the tech. When I went in over lunch on the day of recording, there were 8 people in the room all trying out different things It was great to see so many people taking it seriously and preparing. So it’s all the more demoralising when the usual suspects turned up with no background, in front of cluttered shelves with poor lighting; ignoring every bit of advice we’d been given. The tech team had to talk them through it on the call while 40 others who were all ready to go waited. There will always be some …

My setup consisted of a laptop stacked on two boxes of paper on the dining room table to raise it to eye level, with a bedside lamp on the keyboard shining directly at my face to give consistent lighting.

In the music, there’s an instrumental break; a great opportunity to do something to add a bit of interest and variety for anyone watching. The plan was to use the 10 seconds where the leads don’t sing to drink a beer in a suitably theatrical manner – side on using my full arm to raise the glass. The difficulty being that the virtual green screen doesn’t recognise the glass so tries to cut it out of the picture. Throughout the rehearsals, I’d tried different things to see which worked best and eventually settled on a refreshing beach cocktail – a vibrant yellow drink (orange juice), with a couple of straws which I would have to raise up in front of my body to stop it from disappearing.

The recording
On the day of the recording, I noticed that most other people were wearing sunglasses so made the snap decision to take mine off so I could sing directly into the camera to more directly connect with the viewer – I felt that wearing sunglasses was a way of hiding from the performance and with too many wearing them, it short changed anyone watching. I had to tone down my initial enthusiastic performance given in a previous test run as the sync between the videos meant that any rhythmic movements would be out of time. As a result, my rendition was far more static than I would have liked; it’s a limitation of the technology unless you’re prepared to record each person individually and edit them together – a specialist, time consuming activity.

On the day, we did three recordings, with the third one being the one that was ultimately used, much to my dismay – this was the one where my internet connection slowed, the music became distorted and streamed at an inconsistent pace. It was disappointing for that to happen when I’d put so much into trying to perfect my performance and while I did my best to adapt to these difficulties, my ‘box’ on the final edit is so far behind the others and, to me, stands out like a sore thumb. On the plus side, the video is smooth and undistorted, it’s just hugely delayed! No doubt other people will have had similar experiences, while others won’t have even noticed that anything was amiss.

Ultimately, it was a process that I hugely enjoyed with the right balance of creativity, precision and technological discipline to make it a continuously interesting activity. And of course I received the desired (positive!) comments on taking a drink halfway through. I’m looking forward to the next one!

Chris Shepherd wrote:
It’s worthwhile saying from the off that one of the main things that needs to be bedded in amongst a group of participants on a large video conference is the cultural norm to allow everyone’s voice in the room to be heard. Largely this involves:
– Making sure that everyone is muted unless they are addressing the group
– The leader of the group ensures that they take visual cues from participants who may wish to speak (IE holding up their hand)
– People are encouraged to ask questions by text in the zoom chat channel

Secondly it’s important for technical issues to be resolved by a single point of contact only, and that this generally should happen ahead of time rather than interrupting the flow of the session. One of the key considerations here is to foster a culture where people make preparations ahead of time so they’re not stung by technical issues at the 11th hour.

Finally I’d say time-keeping is crucial. One of the main things I’m experiencing as a software developer is the mental toil that staying on video conferences for 8 hours every single day is having on my ability to function. Allowing time for breaks in any video conference lasting more than 2 hours is essential so as to avoid diminishing returns.

Jim Law was responsible for putting the tracks together. In his account he refers to Joe, who helped with the tracks, and Neil Firth, who is Musical Director.

Jim Lawrenson wrote:
Despite lockdown, I am still working full time from home, so I only really had a few evenings and one weekend to play with, so with only 9 days until the Big Night In, the first step was to move the deadline from Monday to Saturday for the test tracks, knowing it would take some time to work out the process in time for the Tuesday rehearsal.

The first step was to build a test track using as many voices as possible – and then from that, we tried to work out how to build a tutorial for the chorus, that would eliminate the major problems that were likely to arise. Those issues include background noise, clicks, microphone distance and orientation, using different devices to record, misplaced vowels, jumping the beat and lots of others. Auditioning 30 separate tracks of 4 minutes each takes 2-3 hours and that was just to work out which tracks might be usable in the first place, so I realised quickly that it may end up being a fairly time-consuming project.

Using Audacity as a DAW, I brought in a header track for each part, which was referenced to the backing track. Then subsequent parts were imported and matched to the header track one at a time and sync’d and balanced to the header track for that part.

Each sync’d part had to be listened to in it’s entirety with the header track and in some cases multiple times, as the track may have started off correctly, but then changed part way through – so that’s another 4 minutes each track minimum plus processing time for each track – so maybe another 5 – 6 hours.

Although I developed this technique some time ago, it was for a much smaller number of voices, so I was learning about how to do this more effectively and developing the process along the way. In total – it took around 20 hours to get the test track finished. Joe also used a similar process. That’s just to get to the finished test track so it meant some very late nights.

I fed back to the chorus members about how to do a final version on the Tuesday evening, using the learning from the test track and also played it to a few people to get some further feedback. At that point, I reckoned there was about 40% chance of getting a finished version done in time.

The second set of tracks were more in line with our learnings from the test track, so we had eliminated some of the issues, though obviously much of the difference in the recordings is down to the recording device, so some of the tracks had the same issues second time around.

This time, I did quite a lot of work on some of the tracks to edit sections of them – either to silence obvious errors, or to take out serious pops, clicks and other inadvertent noises. In addition, some people started ahead on the early notes, but then got back into sync later on. Rather than just remove the tracks from the recording, I edited each track to silence the start in those cases. That probably took around 5-10 hours.

Once I had the entire chorus lined up and balanced together – I played it to Neil and sent him a few different versions with different levels for each part and he asked to bring out different parts, initially the tenors, to see what difference it made. Then we brought the leads up a bit more and so on.

Once Neil was happy with it, the final step was to incorporate Joe’s version of the track and merge them together.

I realised early on that ideally we needed two versions of the track, either because we could choose between them, or more likely – merge them together to create a much deeper sound. That meant finding someone with a great musical ear and experience of using DAW software. That man was clearly Joe King.

So, Joe did pretty much the same thing that I did, but built a second parallel version of the song using a slightly different method, as he also had some different reverb and echo options – using some of the extra voice filters in Logic Pro X. Sometimes, these can make the sound more robotic, as you will hear on other Virtual Choirs, but as part of a merged track – I figured it would make it sound much better.

We were keen that it didn’t sound ‘processed’ like many of the other virtual choirs, so we merged the two versions to create a 60 voice track to increase the depth of the sound. A bit like Phil Spector’s wall of sound concept.

Another benefit of using two entirely separately created tracks, would be to remove some of the bias in both my ear and Joe’s, and we would end up with a far more balanced final recording. That turned out to be a good hunch as it worked really well.

With the video, I used Adobe Premiere Pro, which is a pretty advanced piece of software, but rather than merge 42 different videos together, Chris figured out how we could record the Zoom meeting in realtime and then I worked out how to post-process it to make it match the audio track as closely as possible.

It’s literally the Zoom online meeting capture, but realigned to the maximum number of voices that are singing words – doots and dums don’t have much mouth movement, so they are less distracting than if someone is singing the words out of sync.

We reckon that between us the finished track took around 50-60 hours, so it isn’t an insignificant project. For future projects, maybe we could halve that if it is a similar type of track. We have looked at testing other types of song – which so far look more difficult to synchronise, so it will likely be back to testing again for those.

Funnily enough, I was looking up the Wall of Sound concept on Wikipedia yesterday – a week after we finished the track and hadn’t realised that Brian Wilson from the Beach Boys, used the same process on the original version of God Only Knows in 1966, so either I think like Brian Wilson, or the song naturally needs that kind of depth to work properly and it’s just a coincidence.

I’m going to go with the first of those.

More thoughts from Steve
(Some of this is repeating what was in last week’s post)
Any time you start using new technology with a large(ish) group of people you are going to come across various problems. One of the first things you need when you are using something like ZOOM to run a group rehearsal and, eventually, a virtual recording, is someone acting as host of the meeting who is infinitely patient and a good communicator. Initially simply getting the whole chorus in the same virtual room at the same time proved to be a problem but those large issues were sorted out within a relatively short time.

The meeting does need a host to manage it but also needs chorus members to show restraint and mute themselves whenever they are not required to be talking. Any extraneous noise can be very off-putting when you are trying to listen to comment or instruction from just one person. Even with a supportive membership, there are always internet links that drop out, people who arrive late, the wrong camera being used on an ipad, audio links that refuse to work or any number of gremlin related hazards to smooth running. Iron out as many of those initial problems and you are in a good position to start.

Our next difficulty was the virtual recording. Because we all needed to be singing the song with the same timing it was necessary to sing the individual parts of the song while also listening to a teach track for that part. The individual tracks could all then be sent off to one of our audio wizards to be assembled into the complete song and leave our only remaining task to mime to that completed song track back on ZOOM. Simple, you might think.

For the miming we decided that “beach attire” and some virtual beach background would be appropriate. The beach attire was reasonably easy. The virtual background a little more problematic. A green screen, or a flat and featureless wall, is a necessity to get virtual backgrounds to work. The singer needs to be well illuminated by a light that doesn’t then cast shadows on the wall or green screen. The wrong lighting or a surface that is not featureless can mean that the singer can disappear into the virtual background each time they move. For some members the only featureless surface they could find was a ceiling. To use this surface the computer/phone/ipad needed to be placed on the floor and the singer needed to lean over it to create an illusion of normality.

It took only four runs through of miming to the song on ZOOM to get a project that “hung together” in a way that we were happy with. A project that we are all, justifiably, very proud of. A project that really brought the chorus together in the lockdown.

April 30 – The total raised is now £1902.
It is not too late to Donate

Leave a Reply

Your email address will not be published. Required fields are marked *