Recording an orchestra

Southae, that’s an excellent post. Thanks for the links and information.

Very cool thread… this is the sort i really enjoy :smiley:

Thanks everyone, lots of great information. I have three ensembles to record: a small string orchestra, a wind ensemble, and a medium sized youth orchestra, around 50 players. I’m now thinking, based on the comments, that x-y makes the most sense, with two LDC’s (cardiod), even though ORTF might make more sense for the largest group. Not to mention it’s my first time! Anyway, still nervous about how to set the levels, given that orchestral music in a live setting has a much larger dynamic range than what I usually record. I guess with 32 bits, I should set the levels lower than I might want to. Any tips on that?

Thanks again, everyone!

My 2c

Decca Tree - good omnis - like Neumann M149’s - over the conducter.

Spot mics like Royers R122’s - say 5.

Millennia pre’s recording @ 96 khz / 32 bit.

Good luck ! It’s the hardest thing to record…BTW

I’ve done a fair amount of recordings of orchestra, string ensembles, wind ensembles, Jazz big bands, etc… This being your first time, the key is to keep it simple. What LDC’s do you have?

Broad thoughts:

Keep it simple - 2 mics as a stereo pair is fine. Generally, SDC’s are a better choice than LDC’s. However, you can still get a very nice sound from a pair of LDC’s. Its probably best to stay away from complex stuff like Decca trees, jecklin discs, lots of spots, etc… Not that these are bad, but they add a degree of complexity that you should avoid on your first outing. Also, in a live setting, you need to minimize things that block people’s sight - which is another reason to avoid trees and discs.

Distance from stage - closer is generally better. The closer you are, the more direct, versus reflected, sound you’ll get. You can’t really take reverb out of a recording, but you can add it quite easily. Placing the mic array 10 feet back from the stage would be a good place to start for a school/youth group. However, you should consider SRA (see below) to refine placement.

Height of mic array - You’ll want some height. Ideally a few feet above the conductors head. This is done for several reasons. First, getting some height helps minimize audience noise. Second, it allows the stereo array to be pointed into the center/back of the group. This helps balance the sound.

Stereo Width (Stereo Recording Angle) - This can be a deep topic, and I don’t want to derail this thread. But you should consider the Stereo Recording Angle (SRA). Here’s the very highly condensed version. Stereo imaging is created by intensity differences (volume differences of left versus right) and/or time differences. Angling directional microphones (such as when using an XY array) provides a stereo image by intensity (volume) differences. Spacing microphones apart (such as when using an AB array) provides a stereo image by timing differences. Spacing and angling microphones apart (such as when using an ORTF array) provides a stereo image by both intensity and timing differences. SRA is an imaginary angle coming from the center of the stereo mic array. Sounds within this angle spread out nicely across the stereo image between a pair of speakers. Sounds coming from outside this angle are pinned hard left or right, meaning that they come from one speaker. You want to keep your orchestra in the SRA, ideally the edges of the orchestra would hit your the imaginary lines of your SRA. Some smart people have figured out the SRAs for various stereo mic arrays. Michael Williams was the pioneer here. His paper, The Stereophonic Zoom, explains this well. You can google for it, go to AES website, or try this Rycote link: https://rycote.zendesk.com/hc/en-us/articles/203908232-The-Stereophonic-Zoom-Archived

You want to avoid making a recording where everything is bunched up in the middle of the stereo field. This happens when picking an array with a wide SRA (like 180 degrees) and then placing the array back a distance from the stage. You want to avoid making a recording where everything is bunched up on the sides. This happens when the SRA is narrow (say 60 degrees) and it’s placed too close to a very wide group. If you want to visualize the SRA for different stereo mic arrays, Sengpiel has a great sight for this. Here’s the link:

Which Stereo Array?
XY - You’ve tentatively picked this one. It will work, but you can probably do better without much effort. Also, XY, where the mics are at a 90 degree angle, produces an SRA of 196 degrees. Very wide. Unless you plan on jamming these mics right up close to the performers, you will have the sound concentrated in the middle of the stereo image and nothing on the sides. Also, XY can sound a bit lifeless and dull. XY is coincident and this will not sound spacious as other arrays can. On the plus side, stereo imaging is good because of the ability of the mics to pick up intensity differences.

AB with omnis - this would be my first choice in a room with great acoustics. It will sound spacious and glorious. However, the downside is this will pick up bad acoustics and highlight them. Although I love this technique, I would not recommend this to you for your first outing. SRA varies depending on how far you space the mics apart.

ORTF - this is accomplished by spacing the mics 17 cm apart with a 110 degree angle between them. Combination of intensity and timing differences for creating the stereo image. Imaging is good, and it has a bit a spaciousness to it. This array, while not always the best sounding one, will almost never sound bad. Bad acoustics? Doesn’t matter - it will still sound good. I recommend you try this one as it is almost impossible to screw it up. It’s a very forgiving array. With an SRA of 96 degrees, it would work well in the situation you noted. Given that you have 3 groups going on and each are probably different widths, consider putting this up within the imaginary angle of the biggest group, then move the mic array closer so you’re clipping the sides a bit. Putting this mic array up would be simple. Imagine that the SRA is actually 90 degrees (easier to imagine than 96 and is close enough) and emanates from the center of the mic array. Move the mics closer/further to stage to tweak your distance as you visualize how wide the group will be. As I noted above, starting out at 10 feet from the stage is probably a good place to start.

Other arrays - there are plenty. Not worth considering for your first time out. Keep it simple.

Safety - If this is a live concert, you’ll want to take a few steps to keep things safe. Bring a roll of gaff tape and tape down your cords (mic cords and any power cords). Use a sturdy mic stand to get the mics up. If your stand isn’t sturdy/heavy, then consider using sandbags to weigh the base of the stand down. People bump into stands at concerts. You can use milk jugs filled with water and a bungee cord as an alternative to sand bags. I use a quickloc A-50 which weighs a ton and has solid steel legs. People have walked into it on several occasions and it has not been knocked down.

Bit Depth - 16 isn’t enough. 24 is plenty.

Sample Rate - 44.1 is fine. You can go higher, but its probably not worth it.

Setting levels - You’ll need some headroom, but you don’t want to leave too much headroom if you can avoid it. Aim for peaks to be between -20 and -12. Leaning towards -20 would be a little safer. If you get a sound check or any time in advance, ask somebody to wack the bass drum for the wind ensemble. That will likely create the loudest peak. Adjust from that. Your levels will need to change for each group. The wind ensemble might be the loudest. String groups tend to be quieter and will likely require more gain.

Backups - Live events generally don’t provide for second takes. If you can, have two machines recording. Use a passive transformer based splitter to make the split. Don’t use Y cords or other active contraptions (these are fine, but not for your first time out). Passive transformer splits don’t need power, automatically isolate the recorders from each other, and are dead simple to use. Even a cheapie splitter like the Art S8 is worth having. Your other choice is to throw up a second pair of mics going into a second recorder. Even putting up a simple Zoom recorder as the backup is worth it.

Hopefully this helps some.

-Tom

1 Like

Excellent post that, thanks Tom :slight_smile:

Tommie-boy, thanks for the great advice. I’ve got a pair of AT2020’s that I picked up for drumkit overheads. Cheapos. I also have four SM57’s, but I do guess that the AT2020’s are more appropriate. Now that you’ve also mentioned ORTF as a better starting point, as Southae did, maybe I’ll go with that. The smaller string orchestra and wind ensemble (around 20 members each) are on one day and the symphony orchestra is the next day (around 50 members). I was worried that ORTF would be too wide for the smaller ensembles, but it is true, they use up most of the stage anyway. Should I do ORTF on both days?

You’re certainly right about having two computers to record, and I certainly wish I had two. I was planning to use 24-bits at 44khz, as you recommend. I may be making a video, but I am pretty sure that the camera compresses the audio to MP3. So I guess that’s my only available backup. Maybe I can borrow a zoom. Do you use one stereo track or two mono tracks in Cubase?

Thanks for the pointers on levels; that’s what I needed. I’ll be able to test it out in dress rehearsal.

Now I have to go do my reading on SRAs!

mbr, what you’re describing is what I’ve read about, but as a volunteer, I’d have to win the lottery first!

:cry:

Hope you win it soon !

I like AT2020s. They are nice mics. I have a pair of the AT4050s, which sound similar. Both the 2020s and the 4050s have a little bit of a high freq bump. This will work in your favor as high frequencies will diminish a bit as sound propagates from the band to your mics.

The 57s wont be of much use. Neither the sensitivity nor feq response of the 57s make them suited to recording orchestra. However, you could set these up as a backup pair for giggles.

I would recommend using ORTF both days. If the smaller ensembles are on one day, consider moving the mics in towards the stage a little more so they fit the SRA better. Doesn’t have to be exact. Your concern about ORTF being too wide doesn’t make a ton of sense when the alternative you were looking at was XY. The SRA is almost twice as wide for XY as it is for ORTF. This is one of those counter intuitive things that people struggle with when they first encounter SRA. You would think ORTF, with it’s 110 degree mic angle would present a wider SRA than XY with mics at 90 degrees. But it doesn’t. Counter intuitively - the wider the mic angle, the smaller the SRA. Also counter intuitively - the further the mics are spaced apart, the smaller the SRA. If you get a chance, fiddle with the Sengpiel link. Start with ORTF and then see what happens if you depart from ORTF by increasing the mic angle (the SRA will decrease). Then change the mic distance and see what happens to the SRA.

By the way, if you don’t have one already, consider picking up an inexpensive adjustable stereo bar. Even something as simple as the K&M 23510 Adjustable Stereo Mic bar (which is only $20). You just need to be able to adjust mic distance and mic angles. If you have a protractor, bring that along too. I picked up an inexpensive one at Lowes. You can use it to set mic angles, and you can use it to help see the SRA.

Question of 2 mono vs 1 stereo file - I don’t track bands with Cubase when I’m in the field. I use dedicated hardware recorders - a Sound Devices 788T is my primary recorder and I use a Roland R-44 for backup. I typically record to one file (poly) then tear the file apart into separate mono tracks when I import to Cubase. If you’re recording directly to Cubase, I might go for a stereo file, rather than 2 separate mono files, as you’re drive is only saving one file at a time, not two. Although your computer can probably handle 2 files fine. If you recorded a stereo file, you can always pull the file apart into 2 separate mono files later if you need to. Panning for ORTF is where the left mic goes hard left and the right mic goes hard right.

If you have a dress rehearsal, definitely record that if you can. It will help you figure out levels. And…if the concert recording has a problem, all is not lost as you at least would have the rehearsal files. Besides - you can have a listen to your recording of the rehearsal and make adjustments for the concert. I have this same scenario with a jazz big band I’m recording in two weeks. I am going to record the rehearsal the night before, then record again at the concert. One of them is bound to turn out OK!

-Tom

Seconding that, really excellent post. Thank you, Tom!

Thanks, Tom. I ordered the stereo bar a couple of days ago. My computer can easily handle 8 tracks at a time, and probably more if I had another input box. I appreciate all the advice you’re giving me!

A few more questions are occurring to me, if anyone can advise.

  1. Do you stop and save the recorded tracks after each number or do you just let it run? I’m on a laptop, and who knows what might make it crash. It never has crashed on me, but my laptop is quite old, running XP, and Cubase 6. I know, it’s due for replacement, but it’s been solid as a rock for recording my band using 8 tracks at once. It has dual disk drives. I think Cubase is doing autosaves along the way.

  2. Processing - I’m thinking basically normalizing, maybe a little reverb if the recording is dry, and equalization to taste. I’m used to recording my band (and I’m pretty happy with it), and I process the crap out of it, so this should be a lot simpler. But maybe it’s standard to do some light compression so that a wild cymbal hit or someone screaming “bravo” just behind the mics doesn’t lower the level of the whole thing?

  3. Presentation - do you give the organization the raw concert from beginning to end? I’m thinking not; I’ll break it into pieces but not down to movements. I’ll capture any conductor comments, and fade out on the applause. If I get decent video, I’ll apply the Cubase audio to the video in Sony Movie Studio, and add a few labels. And I guess nobody will listen to the audio only; they’ll post the video on YouTube, and that’s the final product!

Haven’t read through the current thread completely but did read a lot and asked friends of mine with more experience in this field before my first serious orchestra recording. I’ve been ‘scared’ a little - the kind of being scared when you are to do something you haven’t done before but have to get it right or get in troubles :laughing:

In real life it’s not that hard. There are a thousand confusing methods and opinions but what I’ve found out by myself to get good results is not really much, my personal guidelines (valid for myself, maybe helpful for others):

  • all mics should basically point to the same direction
  • use at least one pair of great mics to capture the whole thing (in my case a pair of heavily modded ADKs), AB/XY does it for me
  • if possible do an additional M/S pair
  • set up spot mics for each group of instruments (like cellos, violins, flutes, horns etc.), at least one/group, if possible more
  • don’t care about not having a huge amount of the worlds greatest preamps (I have a few pretty good, some more standard plus some really cheap in use - nobody ever complained)

Made myself a lot of headaches before the first job of that kind. The more routined I became the more I realized, it’s just Rock’n’Roll too :mrgreen: The stuff I do ends as the audio part of a video production.

Stop, when you are sure it’s safe to stop. Set your pre-capture time high in Cubase (I’m on 9, but I think that feature is pretty old, you’ll find it in the preferences), in case you’ve accidentially stopped you’ll have a buffer.

Probably less is more with orchestra. Basic EQing and some volume rides when necessary. Before anything see what you can do in terms of phase alignment by moving the tracks around a few samples or milliseconds.

Just ask what they want to have.

Thanks for the help, marQs.

What marQ said.

This can be challenging. Depends on the intended use of the recording. If for a competition, you probably should leave it (after bringing whole file up so that peak is around -0.3). If for general consumption (not necessarily audiophile), then I’d bring levels up. Typically, I set markers and then slice the files up by song in Cubase. I then open each song in Wavelab. Step one is to bring up the file so the loudest peak is at something like -0.3ish. Check the loudness. WL does this with the LUFS scale, which is a handy way to look at loudness. You can use RMS as a substitute if you can’t get to LUFS. It’s not unusual for the initial unprocessed file to be in the negative 20s LUFS for loudness. This is just too quiet for most people. I use the loudness normalizer in wavelab to get the file to approx -17 LUFS. This usually makes the song listenable in a variety of environments, but still retains a large dynamic range.

Classical music can be tough to get under control. Depends on the piece. If there is a lot of quiet space and then some massive percussion hits later, you’ll need to consider other options beyond just loudness normalization. First step would be hard limiting to pull the transient spikes of the percussion down. Second technique would be using volume envelopes - this is tricky and should only be done if it can be finessed so as to not be noticeable. Usually a combo of hard limiting and loudness normalization does the trick for pieces that have a wide dynamic range.

My ultimate check is in my car while driving. I pop in a cd (or an mp3 on my phone) of the songs in and turn my volume up half way. If I can hear everything, then it’s good. If not, I consider tweaking. Also, as I’m adjusting loudness, I try to do the adjustment and then listen to the result. If it negatively impacts the piece, then I undo it and try something different.

EQ - Probably not needed with the mics you’re using.

Reverb - need to listen to determine if any is needed. If needed, I lightly use the built in reverb in Cubase - Reverence.

I would break up into songs, rather than having one giant file. Doing fade outs of the applause for each piece is a good idea. I often do fade ins as well if there is room noise that is audible at the start of a song. This way the listener isn’t jarred by the sudden onslaught of room noise.

You’re on the right track!

-Tom

Early21,

On a side note - I think your methodical approach to figuring out things in advance is smart. The first recording I did like this I threw up a pair of 57s in XY at the back of the auditorium. It sounded like utter garbage. Fortunately it was not critical and was just for me and the conductor. I was also manning the conductor’s recorder at the time, which was hooked to two 58s spaced 30 feet apart at the back of the auditorium. His recording also sounded like utter garbage, so I was in good company!

-Tom

Thanks, Tom, these are good ideas. I don’t have the full Wavelab, but I do have Ozone which has many good tools. Meanwhile, I found this article by Mike Senior (who wrote the excellent book Mixing Secrets for the Small Studio):
http://www.soundonsound.com/sound-advice/q-how-should-i-compress-classical-recording

It’s not a recording for a competition so I’ll feel free to experiment.

Interesting read. In the end he mentions parallel compression - I said it before, in the end it’s really just rock’n’roll :laughing: