Skip to main content

MS / spaced omni and phase cancellation technique


I'm working on a holiday Cd with my choirs right now and hope to start recording in 3-4 weeks. I've been happy with using spaced omnis in the fair/poor room I record in and am beginning to experiment with MS to see if that technique will provide better results in the space.

I have always done primarily classical-type recordings of ensembles either acapella or with live instrumentation recorded as one take. This year we're attempting to do a much more "popular style" recording and I will be creating rhythm beds with piano,bass, drums, etc... that we will then overdub the 50 voice choir.

My plan is to use the phase cancellation technique that Zemlin and Remy described here:
(Dead Link Removed)
and that I've done on a smaller scale as a soloist, but my concern is with the phase cancellation in MS and even spaced omnis.

I have the Waves SR1 decoder and I have read enough to know that I will try to minimize high frequencies on the accompanied track but is there any help the pros here can give me, so I don't go down a bunch of paths that I don't really have time to explore and still get this recording done by Mid November. I will experiment but am looking for help.



zemlin Fri, 09/29/2006 - 07:19

I would not expect the record/invert technique to depend much on the mic configuration. It should pickup whatever is coming out of the speakers the same way regardless - as long as conditions are identical on both takes you should be in business.

If your space is less than ideal, I would suggest you also try ORTF (with a cardioid pair). You'll get less room sound than with MS, but can still get a very nice stereo image.

JoeH Sat, 09/30/2006 - 19:20

No offense, but I think you're asking for trouble with the speaker playback/cancellation approach, esp if you want to try out a variety of mic'ing schemes. Way too dicey and complicated, prone to problems, IMHO.

Instead, I'd rent a couple of mulitple-output headphone distro amps like the Roland C-Que8 headphone systems, (or buy 'em outright for about $90 each). I use those boxes for string section overdubs all the time, and it's great; four adjustable outs for four sets x 2 for eight outs.

They take stereo line level in, balanced or unbalanced, and you can run 8 outs, and even use 2-fer splitters for each output, giving you 16 headphone outs per system. Another way to go about it is it to use a low-level FM transmitter and let folks use their own portable FM radio receivers, that will let them have their own volume controls as well. Most folks have their own ipod earbuds that they can use as well, saving you a lot of headsets.

You can also just have every OTHER person use headphones, as well as the conductor, and off you go. The rest can reasonably stay in tune and in pitch, once the song starts.

Just my .02 worth.

Simmosonic Sat, 09/30/2006 - 20:09

JoeH wrote: No offense, but I think you're asking for trouble with the speaker playback/cancellation approach, esp if you want to try out a variety of mic'ing schemes. Way too dicey and complicated, prone to problems, IMHO.

I can understand your concerns, JoeH, but it might be fine so long as a cancellation signal is recorded for each different miking scheme...

My own experience with this technique dates back to the mid '80s, after hearing that was how Robin Millar recorded Sade's vocals for her first album, Diamond Life. She was, in those days, a live performer with little or no studio experience, and so he used that technique to get the best performance. If I remember correctly, he actually got her to silently mouth the words as if she was performing the song again.

Robin Millar was a major influence during my rock/pop engineering days...

Anyway, I tried it quite often with varying degrees of success, mostly with outer-suburban heavy metal and hard rock bands (you know, young guys with hot cars, big guitars, long hair and angst in their pants) because they had very little recording experience and really liked to sing in front of the screaming soffit-mounted Tannoys. Typically, all the other band members would leave the control room (unless the singer wanted/needed someone to perform to, like... the girlfriend!), and we'd get on with it. What we lost in absolute cleanliness of sound we made up for in terms of feel. It also sped up vocal sessions enormously; there always seemed to be less mistakes, and the communication between engineer and artist was great.

In the late '90s I was talking to engineer/producer Rickster Will, and he was using the same technique to record Jimmy Barnes belting out his vocals in the control room.

I can say it works well in a close-miked studio environment. I am not sure how it will go in a larger space with reverberation and so on.

Coincidentally, I have a similar choral job coming up in late October: the choir did a live performance and had it filmed, but the numbskull camera guy forgot to plug in the microphone (that's the excuse I was given) and so now they want to 'reverse mime', re-recording the sound to fit the silent video footage. Fortunately, they performed to a pre-recorded musical accompaniment, so the plan is to play the accompaniment back, sing the piece again, record it, and dub it onto the video. It's a 50-voice amatuer womens' choir, and most are housewives with no experience of singing with headphones on, so the headphones idea just isn't gonna happen!

Simmosonic Sat, 09/30/2006 - 20:27

zemlin wrote: I used the technique in an elementary school gym recording 150 5th graders singing to a karaoke CD. I had one pair of ORTF mics. It worked quite well.

Ah, thanks Zemlin! That's the kind of thing I was hoping to hear, because the recording I've got coming up is happening in a community hall and I'll almost certainly be using ORTF... If it can work with 150 school kids in a gymnasium, it can probably work anywhere. :)

zemlin Sat, 09/30/2006 - 21:03

Here's a quick demo of that session.
45 seconds - the first 15 seconds are as recorded with speaker bleed.
The next 15 seconds mixed with the music only take inverted.
The final 15 seconds is with the original music mixed back in.

Keep in mind, this music only take was done without the singers in place - they simply couldn't have been quiet for that long. I did some test takes in the same space, and when conditions were the same for both takes, cancellation was excellent.

[edit] what 'da heck ... here's one of the tests [/edit]

pmolsonmus Mon, 10/02/2006 - 07:41

JoeH wrote: No offense, but I think you're asking for trouble with the speaker playback/cancellation approach, esp if you want to try out a variety of mic'ing schemes. Way too dicey and complicated, prone to problems, IMHO.

Absolutely no offense taken. I have 2 furman headphone amps that I've used with my vocal jazz groups. (16 outputs but need to be close to the amp or buy add'l cable for choir work- willing to try if worth it). With vocal jazz we use a much closer group mic (omni or cardoid or fig8) The problem with this technique with a 50 voice choir is the mics will be much further back and there's no real way to have the students hear their voices through the headphones and then we end up with bleed or blend/ balance issues with the students not wearing headphones. 16 students with 1 can off is going to be a high frequency nightmare for any stereo technique. Yes??????

Please keep ideas coming. I'm also considering what Johnny Mandel did with Shirley Horn's "Here's to Life" Cd. Record the vocals with very light piano comping and then add more piano/bass/drums/strings etc... over the top. Thoughts??? The instrumentalists will be pros. I could make a quick practice CD with the vocal takes and give them a copy. Is this a better route?

Zemlin, don't get me wrong. I like the idea, and am willing to try but the choir has to sound great. If they don't sound great they won't want to sell the CD and the Cd is to help pay for a choral festival at the Kennedy Center in DC in spring. This choir can sing and sing well. The whole point in going "popular" is to be profitable for this trip.

Simmosonic Wed, 10/18/2006 - 02:18

zemlin wrote: Here's a quick demo of that session.

Your demonstration recording of the cancellation is very impressive; all the problematic low frequencies (large wavelengths) are gone, leaving only some of the high and mid frequencies which I doubt would be much of a problem. The technique works pretty much as well in a larger space as I recall it did in the studio.

The excerpt from the finished recording is delightful, and I doubt the client would be complaining about spill from the backing music!

By the way, I'm a real sucker for groups of kids singing with such joyful abandon. When I'm travelling in the Himalayas or Asia and come across a school and/or a group of kids, I always ask if they'll sing a song for me to record.

[Please accept my apologies for taking so long to reply/respond to this...]