4

I direct a school-aged chorus of 100 students, divided equally into sections A and B. The two sections sing different parts of 2-part arrangements. Our COVID restrictions this year state that: (1) I cannot have more than 50 students in the rehearsal space at once, and (2) I can not have students from different section in the space at the same time. So I either have all of A, or all of B, for each rehearsal. Further, I am not permitted to hold any live, in-person performances. Even with all this in mind, the students have enjoyed singing and want to share what they are capable of. I'm planning to produce a recording of their singing and am planning to try it like this...

I have at my disposal enough devices (chromebooks, ipads, etc) to supply each student with one. Here is my plan. I'm looking for ways that this may fail before I try it with 50 kids.

  1. On Day 1, Record the accompaniment only with my accompanist and me in the empty rehearsal space.
  2. On Day 2, invite Section A into the rehearsal space. I'll host a Zoom session, invite all the students, mute them, and instruct them to put on headphones (we have enough). From the device hosting the Zoom, I'll play the accompaniment file. They will all hear it in their headphones and be able to sing along. If I also listen to the file as a participant in the zoom, not the host, I should experience exactly the same latency as them when I conduct, and so we will all be in sync and hearing the accompaniment at the exact same time. I'll use a separate device to record their singing.
  3. On Day 3, I'll repeat this process with Section B.
  4. After the recording days, I'll use ProTools or GarageBand to mix the accompaniment, section A, and section B together. This will not result in a perfect studio recording (obviously), but it should be a reasonable "capture" of their performance.

In what ways is my strategy most likely to fail, has anyone tried anything similar, and are there any additional strategies that I might be able to employ?

9
  • 2
    Based on one of my project managers complaining twice that my audio output is breaking up over Zoom when I can hear them just fine on my end in the same meeting, I bet at least one but not all of your students will end up in a similar situation.
    – Dekkadeci
    Oct 13 at 12:37
  • 22
    "I should experience exactly the same latency as them when I conduct," Based on my experience, that's not a valid assumption.
    – Duston
    Oct 13 at 13:12
  • Quick question: I assume the repertoire would not be suited to a metronomic "click track"? If you want organic rubato, that option is out. But if not, even tempo and meter changes could be programmed in, and synchronization of recorded sessions could become much easier. Oct 13 at 13:17
  • 1
    Recording people separately is good; using Zoom I doubt would work out.
    – JDługosz
    Oct 14 at 15:49
9

As existing answers & comments have established, forget Zoom. Also forget trying to run 50 sets of headphones through an amp - as far as I know 8 channels is luxury in headphone amps, unless you have a language lab you can borrow one from.

However, if your 'backing mix' is in mono & you use a pair of speakers with flipped phase on one side [literally just swap the cables +/- going to one speaker], then you may be able to reasonably 'hide' the leaked audio by putting the mic[s] centrally between speakers. I've never tried this myself in a large space or in a choir setup*, but it works very well for a single hand-held vocal.

One advantage is you could spend as much time as you like before the session testing out variations, see what works before you get the kids in.

*BTW, if you're trying to mic the choir in any kind of stereo configuration, then set the speakers front & back, then your mics can be stereo to the choir but central to the speakers.

3
  • 2
    The stereo-cancelling trick is a great idea, but I'm afraid it probably won't work well at all in a big room with significant natural reverb. What should be possible (even with one speaker) is to once record the return from the backing track whilst the students are quiet, then once with them singing, and mix both tracks with opposite phase in the end. Oct 14 at 7:30
  • @leftaroundabout both phase switching ideas sound to me like something that should technically work, but doesn't actually work with natural reverb, etc. Does it actually work?
    – nuggethead
    Oct 14 at 9:49
  • @nuggethead I haven't tried it myself, but yeah: as long as the speakers & microphones stay in the same place, all the reflections will be the same on both recordings, and thus should cancel in the end. It's won't be perfect, but if it's just the backing track then it should be plenty good enough. If there's also a click track, I wouldn't risk it. Oct 14 at 9:54
7

I definitely have my doubts about the approach. (Also, TLDR spoiler: It's not really a new problem, and there are other solutions.) Among the problems a pessimist could imagine:

  1. On recording day, you open yourself up to "My device isn't working/ran out of batteries/my headphones aren't working/I'm having trouble with the wifi" etc. times 50. The odds that the technology co-operates smoothly are slim, and rehearsal time will probably be lost.
  2. The assumption that everyone's latency will be equal is not a given. It can be affected not just by internet connection, but (I imagine) by issues local to each device.
  3. 50 devices all trying to access what is probably the same router might cause their own issues (disparate latencies, or even a traffic jam). They might even create issues within Zoom's logic handling a 50-person call.

But ultimately this isn't a new problem. Any time anyone wants to add a choir or orchestra to an existing track, they face the question: How do I provide monitoring to a large ensemble? How can I help them stay in time with the existing track?

Unless you have the ability to provide in-ear monitoring to that many people, the solution is normally to just give the conductor headphones and trust the ensemble to follow them. With the disclaimer that I'm a performer, not a recording engineer, I imagine I'd:

  1. Record the accompaniment. I might even treat this as a "scratch track," to be replaced later. If there are complex rubato or tempo changes, I might even record myself counting out loud so that I can synchronize my later conducting, preferably in a way that isolates this from the piano track.
  2. On recording day, I would try to provide the piano track in normal speakers so the choir has its support for intonation and timing. I would have my own counting in-ear and would conduct.
  3. If everything syncs up well, great; if necessary I might do a "final recording" of the accompaniment to accomodate any fluctuation that happened in the choir recordings.

I seriously wonder whether it's better to record part A and part B separately... or to do half of A with half of B, then the other halves? There are obvious pros and cons. In a sectional, each part can hear itself well. But they're missing cues from the other part and can't tune against it. And I wonder whether which method would yield more problems in synchronization...


Update: Oh, and I'd definitely advise doing the whole project, from the start, in your DAW, ProTools or Cubase or even GarageBand. If you were to try something like "Record Part A on some digital audio recorder, record Part B, then try to sync them up in post," that's a tricky business. Any concern about synchronizing the bleed-through of the accompaniment will be minimized if the DAW can take care of the relative timing of the tracks; then presumably the two copies of bleed-through will be perfectly aligned and not be a problem.

I guess this makes my suggestion of treating it as "scratch piano" a bad one. Maybe shoot for finished piano right off the bat. And if you count out loud for your own benefit, this would have to be fully isolated from the piano. Another approach: If your DAW supports video, you could capture video of yourself conducting at the same time as recording the pianist, then reference this when conducting the choir.

13
  • OK @Andy Bonner I think you've scared me off of relying on Zoom! My hesitation with providing live monitor speakers is that the accompaniment would be audible in the recording ... AND when I mix Part A and Part B together, it will be present TWICE in the recording and cause audio issues there. But that may be preferable to the syncing /wifi issue. Careful mic/speaker placement could mitigate it, I suppose. Thanks for the answer!
    – nuggethead
    Oct 13 at 14:11
  • Unfortunately, I can't allow A and B to be present at the same time for rehearsals. Logical or not, it's our covid policy and not something I have any say over. Yes, this makes it very hard to sing their part in tune and blend/tune with the other part, but such is life.
    – nuggethead
    Oct 13 at 14:14
  • @nuggethead Yeah, I share the concern about speaker-monitoring bleeding into the recording. I hope somebody with more recording experience than me can speak to that, because surely it's a common problem—either you provide in-ears to all the members or they "sing blind"! Oct 13 at 14:32
  • 1
    @GlennWillen Yes, that's of course why movies use clapperboards in the first place. My personal trauma comes from recording using a video camera and an H4N audio recorder simultaneously, then wanting to replace the video audio with the H4N audio, and using nothing more powerful than some kind of Windows Movie Maker thing that came with Windows ME, nudging tracks forward or backward milliseconds. If I'd had a rig that could just capture the video and audio simultaneously, it could have saved some tears! Oct 13 at 20:47
  • 1
    @AndyBonner: Synchronization of audio and video is complicated by the fact that the change from NTSC to digital video did not result in an adjustment of the frame rate from 29.97 frames/second to 30. Tweaking the video frame rate from 30fps to 29.97 may have seemed a clever trick to avoid the interference that would have been caused by using a chroma subcarrier that's 119,437.5 times a 30fps frame rate, but it's cursed video editors ever since.
    – supercat
    Oct 13 at 22:25
6

Horribly complicated and prone to glitches.

Do it this way. Record the accompaniment. Point some speakers at the singers. Point some microphones at the singers. As far as possible, try not to point the microphones at the speakers!

Yes, you'll get bleed. But you WANT the accompaniment in the recording :-)

Maybe you could wire everyone up to headphones (wired, not via Zoom). This is how a pro studio might do it. But that's a LOT of headphones! And think of having to sanitise them before the next group used them (COVID, you know).

It'll be fun. Especially if you don't let the technicalities get in the way of making music. Remember this is just a stopgap project until you can meet properly again. The aim is participation, not a top-quality recording! (Though you may be surprised how well a relatively low-tech approach works.)

3
  • With speakers and directional (e.g. supercardioid characteristic) microphones this can work, and might be the best solution available at low budget. Oct 13 at 20:26
  • 1
    Also, if the recording space isn't treated for reflection control, consider hanging something soft and fuzzy like a blanket on the wall behind the singers. That should both reduce bleed and make the recording drier.
    – Dan Bryant
    Oct 13 at 21:28
  • We're talking about 50 singers here! That's going to be a BIG blanket! But yes, if this is being done in a school hall, maybe there are stage curtains... Oct 17 at 10:19
0

Here is your principal mistake:

I should experience exactly the same latency as them when I conduct, and so we will all be in sync and hearing the accompaniment at the exact same time.

You'll all be hearing the accompaniment at the exact same time and will be "in sync" to an external observer, but your conducting will appear too late for your choir members and their singing will appear too late both for their peers and for yourself. In addition, conference software does ducking: when you speak, it will lower the volume of the other side. That makes for rather weird choral performance.

It would probably be better if you got people onboard with trying out Jamulus and got a good server for yourself. Jamulus is built to minimise latencies for musical cooperation. I've used it for working with small ensembles and it gets latencies down comparatively well, given sensible connectivity of the participants and the server. Most importantly, the bad effects of one person's less than optimal connection are localised: they have to deal with their own roundtrip latency (which they hear in the mix). Some practice helps though it's likely harder for singers than for keyboard players to adapt to such latencies.

Short of a setup/application that allows a group experience, basically you are down to sending around your backing track and let people record themselves to it, collecting the results.

A mixture would be to create a base track with a small set of people and let others work offline based on that.

All of this will mean considerable effort, but you won't get anywhere really with a group Zoom session.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.