OleJoergensen
Senior Member
Thank you for sharing this knowledge Mike.
The Adventures of Creating a Choir Library - April 26, 2024 Update - Wait ... we've spent HOW MUCH???
- Thread starter Mike Greene
- Start date
DSmolken
Senior Member
Not to make this about me, but yeah, choir libraries sure can involve messy logistics. Hadziha has two sets of vowels recorded with two sets of singers because after starting the recording one singer moved to a neighboring country, the whole thing died, and it took months to recruit new ones. But hey, it sure was worth it, it's probably sold about as well as the solo strings and orcs, and definitely much better than the solo voices.
Hopefully things go smoother for you, and nobody loses their voice three days in or anything like that.
Hopefully things go smoother for you, and nobody loses their voice three days in or anything like that.
angeruroth
Meeb Avemcrit
Hey! Just chiming in to say thanks 
I'm one of those always interested in this kind of thing, and to see it with real data included, well, what can I say? I'll follow this thread with utmost interest.
About the shorts, I have to say that the one thing I miss in some textural libs (and at this point I use textures everywhere, all the time) is precisely that.
To give an example of why, I still have to find the right shorts to match the sound, the space, etc. of BDT. And I love that lib, but unless you go directly for shorts with a different sound it can be really, really frustrating. I mean, you can render a counterpoint with a different lib and it can sound good even if the size is completely different, but it's not the same, so to have some shorts included, especially if one thinks about the texture of those shorts, is not only great but sometimes almost a necessity IMHO.
I'm one of those always interested in this kind of thing, and to see it with real data included, well, what can I say? I'll follow this thread with utmost interest.About the shorts, I have to say that the one thing I miss in some textural libs (and at this point I use textures everywhere, all the time) is precisely that.
To give an example of why, I still have to find the right shorts to match the sound, the space, etc. of BDT. And I love that lib, but unless you go directly for shorts with a different sound it can be really, really frustrating. I mean, you can render a counterpoint with a different lib and it can sound good even if the size is completely different, but it's not the same, so to have some shorts included, especially if one thinks about the texture of those shorts, is not only great but sometimes almost a necessity IMHO.
QuiteAlright
Gastronomer
Thanks for writing about this. I'll certainly follow along as you write more about your progress, and I think this is a super interesting topic.
How do developers like yourself pick the performers to sample? Or in other words, do you already know some singers who you intend to hire? How do you go about finding people, and vetting them, for a project like this?
I'm also curious about how much time editing and programming take for a project like this versus time.taken to record the sounds in the first place.
How do developers like yourself pick the performers to sample? Or in other words, do you already know some singers who you intend to hire? How do you go about finding people, and vetting them, for a project like this?
I'm also curious about how much time editing and programming take for a project like this versus time.taken to record the sounds in the first place.
Geoff Grace
Senior Member
I have a placeholder response until Mike is able to answer. My experience is that you find a contractor you trust, and the contractor books the singers/players. LA has a wealth of talent. There are usually very good musicians available when you need them.
Alternatively, you can book an established ensemble, as Spitfire did with Eric Whitacre.
Best,
Geoff
kgdrum
“Making a mess & calling it Art is where it’s at!”
@Mike Greene
“Have you ever wished a long-winded, self-indulgent egomaniac?”
Of course it’s not just you being incredibly handsome,some of us really enjoy long winded posts from suave narcissistic developers!
“Have you ever wished a long-winded, self-indulgent egomaniac?”
Of course it’s not just you being incredibly handsome,some of us really enjoy long winded posts from suave narcissistic developers!
Last edited:
NYC Composer
Senior Member
But…but….I want the quintet!
My lack of a great string quartet (or quintet) has driven me mad.
My lack of a great string quartet (or quintet) has driven me mad.
OP
OP
Recording starts next week, so in this post, I’ll talk about what the plans are.
THE DAILY SCHEDULE
Sunset Strings was a section library, where the violins, violas and cellos were all recorded together. Personally, that’s how I like to play a library, since it’s instant gratification, plus Sunset is a textural library, not a tool for mocking up your classical masterpiece. (We also included basses, but those were recorded separately, so the user could decide whether they wanted them or not.)
So we decided to do the same here, although we’re recording men and women separately. Basses and tenors together, and altos and sopranos together.
The sizes of the sections was the big question. It was tempting to go with an intimate and small choir (12 altogether, so 3 in each “section”), but that makes me nervous on whether a choir that small will be “pretty” enough. So we decided to go with 20. (Okay, actually it was just me that was nervous about having only 12 singers, but … being the guy who writes the checks does have its privileges.)
So here’s how it will work - For the men, the five basses will sing the lowest notes and the five tenors singing the highest notes. We decided not to have all 10 sing the middle notes, though, because we don’t want it too thick. So we’ll be having maybe 3 basses and 3 tenors through the middle section. That will probably be a “by ear” decision we make as we hear how things sound on Monday. We might also taper the transition for 5 basses to 3basses/3tenors and maybe do transitions of 5basses/1tenor, then 4basses/2tenors on the intermediate notes. Seems like the obvious thing to do, but it would complicate things (complications = bad!), so we’ll probably play it by ear.
The contractor tells me the singers can do about four hours in a day, so we’re doing women for four hours in the afternoon, then men for four hours at night. With a one hour dinner break, that’s nine hours. A studio lockout is 12 hours, and I’m the kind of guy who wants to get maximum value for his dollar, so we’ll add a three hour 3-singer session each morning, with basses one day, tenors another day, etc. With lunch break, that’s a 13-hour day. That puts us into overtime, but we gotta do what we gotta do.
Those morning sessions will focus mainly on the “Repetitions,” which is where we record singers solo doing various things individually. With the strings, we had them do ricochet’s, for instance, where we recorded a whole bunch of round robins of solo ricochets at each note, which we could then control via the interface how dense the ricochets would be as they’re mixed in with one of the sustains. We did this because the original (and more obvious) plan was to have the section play a sustain as random players play ricochets at random times. Cool, but there’s no control if we did it that way, so we made the ricochets (and all the repetitions) a separate thing that you can control. (I’m really proud of “our way” on this, and in fact, that’s what prompted Sunset to be a two-layer instrument, since you’ll generally want those ricochets to be playing along with a sustain.)
So the morning 3-hour sessions will be solo singers, while the main 4-hour sessions in the afternoon and evening will be mostly section singing. The main reason for that is that if we did the individual singer notes in the afternoon or evening sessions, then nine singers are standing around doing nothing while one singer sings “bop” or whatever little repetition ideas we’ll try. At almost two grand an hour, you don’t want that. So a separate session in the morning with just three singers will be more efficient.
THE RANGES
Deep, deep basses and stratospheric sopranos seems the obvious goal, but there are a couple things to bear in mind. One is that extended ranges means more samples, which means more recording time. More recording time per articulation means fewer articulations.
It also means stressing the singers’ voices, which I don’t think is worth it. Similarly, I don’t think this library will calls for fff dynamics, so we’ll probably skip that.
I checked out some other libraries and the Spitfire Eric Whitacre library sounds great, and seems to be accepted as a top notch library, so … why reinvent a wheel they already invented? So we’ll probably just copy them and go with these ranges:
Basses - E1 to Db3
Tenors - B1 to G3
Altos - E2 to Eb4
Sopranos - B2 to A4
ARTICULATIONS
We had lots of ideas going into this, but we also listened to a lot of soundtracks and pulled ideas from those as well. Those are the easiest, because we can play an mp3 snippet at the studio, and I can say, “Do that!” It should be fun, because they’ll likely enjoy doing “weird stuff.”
We’ll try to do some legatos, but we’ll see. Legatos are soooo time consuming, although I do have some tricks. If we do that, the emphasis will be on slow legatos, like the “Slow Bends” in Sunset Strings. It’s an unusual, but useful articulation, so we think it would be cool.
Choosing which vowels to record is going to be tricky. Obviously we’ll want oo, ah and mm. I’m thinking oh and uh would also be useful. The problem is that unlike strings, if we record attacks (slow bends, crescendos, or whatever), we need to do versions for all the vowels. Or maybe we don’t, and just say certain attacks/releases only work with oo and/or ah. This is the sort of thing we’ll see as we go, and it’s why I’m reluctant to define what the instrument will actually be.
Another thing we’ll see as we go is whether we record dynamics. The temptation is always yes, but bear in mind that sampling is a multiplication exercise, where two dynamics means double the recording time. Same with round robins. Or recording chromatically instead of by whole tones (which seems to be how everybody records their choir libraries.) Record two dynamics, two round robins, and do it all chromatically and it will take eight days to record what you could have recorded in one if you stuck with one dynamic, no round robins and sampled by whole steps. Math can be so cruel!
THE VARIABLES I WONDER ABOUT
Recording Sunset Strings was an incredible experience, because the players were so good. It truly was amazing to have a front row seat as they played. (I know I promised in the first post not to gush, but the sound, being right there in the room, was amazing.)
The most pleasant surprise with the strings was their tuning. We accomplished a lot more than I expected in the recording sessions, because we didn’t have to make as many tuning corrections as I expected. In fact, at one point they told me I didn’t need to keep playing the next note on the piano. “We’ll get it. Trust us.”
It will be interesting to see if singers have the same pitch command that string players do. Not only do they not have frets, they can’t even see where their finger is on the neck! I’m guessing tuning won’t be an issue (these ain’t cheap singers!), but I still can’t help but worry.
There’s also some question about how long they can hold a note comfortably. I don’t want to push those limits, because if they’re consciously conserving breath, it will affect tone. Doing some tests myself, I think we’ll go for 8 second sustains. Since this will be a textural library, with movement as opposed to static notes, we don’t want short loops.
Continued in next post ...
THE DAILY SCHEDULE
Sunset Strings was a section library, where the violins, violas and cellos were all recorded together. Personally, that’s how I like to play a library, since it’s instant gratification, plus Sunset is a textural library, not a tool for mocking up your classical masterpiece. (We also included basses, but those were recorded separately, so the user could decide whether they wanted them or not.)
So we decided to do the same here, although we’re recording men and women separately. Basses and tenors together, and altos and sopranos together.
The sizes of the sections was the big question. It was tempting to go with an intimate and small choir (12 altogether, so 3 in each “section”), but that makes me nervous on whether a choir that small will be “pretty” enough. So we decided to go with 20. (Okay, actually it was just me that was nervous about having only 12 singers, but … being the guy who writes the checks does have its privileges.)
So here’s how it will work - For the men, the five basses will sing the lowest notes and the five tenors singing the highest notes. We decided not to have all 10 sing the middle notes, though, because we don’t want it too thick. So we’ll be having maybe 3 basses and 3 tenors through the middle section. That will probably be a “by ear” decision we make as we hear how things sound on Monday. We might also taper the transition for 5 basses to 3basses/3tenors and maybe do transitions of 5basses/1tenor, then 4basses/2tenors on the intermediate notes. Seems like the obvious thing to do, but it would complicate things (complications = bad!), so we’ll probably play it by ear.
The contractor tells me the singers can do about four hours in a day, so we’re doing women for four hours in the afternoon, then men for four hours at night. With a one hour dinner break, that’s nine hours. A studio lockout is 12 hours, and I’m the kind of guy who wants to get maximum value for his dollar, so we’ll add a three hour 3-singer session each morning, with basses one day, tenors another day, etc. With lunch break, that’s a 13-hour day. That puts us into overtime, but we gotta do what we gotta do.
Those morning sessions will focus mainly on the “Repetitions,” which is where we record singers solo doing various things individually. With the strings, we had them do ricochet’s, for instance, where we recorded a whole bunch of round robins of solo ricochets at each note, which we could then control via the interface how dense the ricochets would be as they’re mixed in with one of the sustains. We did this because the original (and more obvious) plan was to have the section play a sustain as random players play ricochets at random times. Cool, but there’s no control if we did it that way, so we made the ricochets (and all the repetitions) a separate thing that you can control. (I’m really proud of “our way” on this, and in fact, that’s what prompted Sunset to be a two-layer instrument, since you’ll generally want those ricochets to be playing along with a sustain.)
So the morning 3-hour sessions will be solo singers, while the main 4-hour sessions in the afternoon and evening will be mostly section singing. The main reason for that is that if we did the individual singer notes in the afternoon or evening sessions, then nine singers are standing around doing nothing while one singer sings “bop” or whatever little repetition ideas we’ll try. At almost two grand an hour, you don’t want that. So a separate session in the morning with just three singers will be more efficient.
THE RANGES
Deep, deep basses and stratospheric sopranos seems the obvious goal, but there are a couple things to bear in mind. One is that extended ranges means more samples, which means more recording time. More recording time per articulation means fewer articulations.
It also means stressing the singers’ voices, which I don’t think is worth it. Similarly, I don’t think this library will calls for fff dynamics, so we’ll probably skip that.
I checked out some other libraries and the Spitfire Eric Whitacre library sounds great, and seems to be accepted as a top notch library, so … why reinvent a wheel they already invented? So we’ll probably just copy them and go with these ranges:
Basses - E1 to Db3
Tenors - B1 to G3
Altos - E2 to Eb4
Sopranos - B2 to A4
ARTICULATIONS
We had lots of ideas going into this, but we also listened to a lot of soundtracks and pulled ideas from those as well. Those are the easiest, because we can play an mp3 snippet at the studio, and I can say, “Do that!” It should be fun, because they’ll likely enjoy doing “weird stuff.”
We’ll try to do some legatos, but we’ll see. Legatos are soooo time consuming, although I do have some tricks. If we do that, the emphasis will be on slow legatos, like the “Slow Bends” in Sunset Strings. It’s an unusual, but useful articulation, so we think it would be cool.
Choosing which vowels to record is going to be tricky. Obviously we’ll want oo, ah and mm. I’m thinking oh and uh would also be useful. The problem is that unlike strings, if we record attacks (slow bends, crescendos, or whatever), we need to do versions for all the vowels. Or maybe we don’t, and just say certain attacks/releases only work with oo and/or ah. This is the sort of thing we’ll see as we go, and it’s why I’m reluctant to define what the instrument will actually be.
Another thing we’ll see as we go is whether we record dynamics. The temptation is always yes, but bear in mind that sampling is a multiplication exercise, where two dynamics means double the recording time. Same with round robins. Or recording chromatically instead of by whole tones (which seems to be how everybody records their choir libraries.) Record two dynamics, two round robins, and do it all chromatically and it will take eight days to record what you could have recorded in one if you stuck with one dynamic, no round robins and sampled by whole steps. Math can be so cruel!
THE VARIABLES I WONDER ABOUT
Recording Sunset Strings was an incredible experience, because the players were so good. It truly was amazing to have a front row seat as they played. (I know I promised in the first post not to gush, but the sound, being right there in the room, was amazing.)
The most pleasant surprise with the strings was their tuning. We accomplished a lot more than I expected in the recording sessions, because we didn’t have to make as many tuning corrections as I expected. In fact, at one point they told me I didn’t need to keep playing the next note on the piano. “We’ll get it. Trust us.”
It will be interesting to see if singers have the same pitch command that string players do. Not only do they not have frets, they can’t even see where their finger is on the neck! I’m guessing tuning won’t be an issue (these ain’t cheap singers!), but I still can’t help but worry.
There’s also some question about how long they can hold a note comfortably. I don’t want to push those limits, because if they’re consciously conserving breath, it will affect tone. Doing some tests myself, I think we’ll go for 8 second sustains. Since this will be a textural library, with movement as opposed to static notes, we don’t want short loops.
Continued in next post ...
OP
OP
Continued from previous post...
A COUPLE NOTES ON HOW WE’LL RECORD
Typically, for each articulation, you start at the lowest note, then go up the scale for the full range, then move on to the next articulation. I came up with a different technique as I was recording the Realivox Ladies, though. I start with a note in the middle, then go up from there, then come back to that middle note, and then go down from there. Sort of a “Middle Out” thing. I do that because by starting in the middle, I can be sure I’m getting what I like to call “the money tone” and make adjustments if not. If you start on the lowest note, you’re in outlier territory, not really hearing what the actual articulation is going to sound like in the sweet spot until you actually get to the middle notes and realize, “Oh, this is too dark/bright/nasal/whatever!”
We did that with Sunset Strings, where it also had the advantage that since Sunset was recorded as an ensemble, it’s only the middle notes where all the cellos, violas and violins played. We’ll probably do the same thing with the choir.
I mentioned earlier that the string players could find pitch for each note without needing me to play a piano note for them. They also didn’t need to play a short tuneup to make sure their finger was in the right spot. (With no frets!)
My experience with singers is that they’re not that quick to center their pitch for each successive note. (Even more so than with instruments, recording a vocal sample library is a completely unnatural process, since singers don’t sing straight notes in real life.)
In fact, with some of the Realivox Ladies, on some of the samples, you can hear a slight ramp up at the beginnings of the notes. That’s because if they’re singing an F, then that means they had just sung an E right before that. So they still have some of that E in their head, and in the muscles of their vocal cords, so there’s that tiny little slide into the F.
So I have a new idea I might try out with the choir. With some of the basic articulations, rather than doing the articulations one at a time, where we sing a note, then sing that same articulation on the next note, I might instead focus on one note at a time. In other words, do a bunch of articulations at that one note, so their pitch will already be centered as they sing oo, then sing ah, then sing oh, all on that same note. Then go to the next note and do the same sequence. The risk of that is we may lose consistency on what exactly an oo or ah sounds like, but we’ll see.
That will only work for the basic articulations, of course, where they already know what oo and ah are, without me explaining them. For the unusual articulations, we’ll have to do those the normal way.
THE MECHANICS OF IT ALL
We make a chart (spreadsheet) of all the things we might do. We add numbers at the front of each articulation name, so they’re easy to find. For instance, we might have “01 - Oo Sustains” and then “02 - Oo Legatos.”
A single massive PT project would get unwieldy, so each of those numbers will be its own Pro Tools project. Then again, 200 separate Pro Tools projects would be a mess, too, so we’ll probably compromise with something like “01 - All Vowel Sustains” and the “02 - All Basic Legatos.”
Keeping track of what’s been recorded (and what hasn’t) gets very complicated. This is not a typical library, where you have a definitive list of what needs to get recorded. We’ll try to have a list of everything we can think of, but I guarantee we’re going to hear one of the singers doing some weird warmup that will give us ideas for more articulations.
As we add things to the list (and … cross things off! Yikes!), we’ll need to keep an eye on whether the stuff we’re keeping will make for a cohesive product. So we’ll probably be re-prioritizing what we *need* to record as we run out of time on days four and five.
Vincent will be at the sessions, and since he’s involved in the pre-production, he’ll probably be involved in managing that. Karen (our contractor) also does that sort of thing, since she’s also at all the sessions. We’ll also have Gabriel, although he’s not as familiar with the project. (He’s my nephew, and works at the studio doing other projects. Talented, but not an orchestral guy.)
Jayden is in Australia. He’ll probably be watching via Zoom, but his day is our night, so we’ll see. (Our normal Zoom meetings are at 4 in the afternoon, which is a good compromise time.) The plan is to “Audio Mover” the stereo roughs as we record, so he can make some basic patches (while we’re asleep) and let us know the next morning how things are working. That will help determine which paths we continue on, and which paths we pull the plug on.
MEET YOUR CONDUCTOR!
One of the nice things about writing the checks is you get to give all the fun jobs to yourself, and the funnest job of all is conducting. It’s stressful, but I like doing it.
Conducting the strings a couple years ago was entirely new to me, but you don’t exactly need a PhD in Conducting to move your finger in the 1-2-3-4 pattern. (I did buy a baton, but I used my finger instead.) This is mostly long sustains, so it ain’t rocket science. (Well, the full project is sorta rocket science, but the conducting isn’t.)
I’ve recorded a zillion singers over the years, and I like to believe I’m pretty good at it, partly because one of my strengths is in keeping the singer’s energy up. (With singers especially, that’s important.) There’s a whole psychology to it, where I try to stay upbeat as much as possible, because their energy feeds off of mine. (Staying upbeat for 13 hours is impossible, so I’m also hiring a real conductor to fill in for me through the day.)
Another advantage I have is that I’ve learned how to ask for re-takes in a positive way. Instead of “You were out of tune on that one. Do it again.”, I might say, “Your tone was great! Let’s do one more for tuning.” Or the old, “Let’s do one more for a safety” trick. I’ll do whatever I’ve got to do to keep them motivated and engaged. (Mind you, it’s not all trickery, and sometimes I do say simply that they were out of tune, but I’m watching for how they’re receiving that note.)
Most importantly in the “Why I conduct this myself” list, is that with this project (as with Sunset Strings), I’m the guy who knows exactly what we’re trying to accomplish. I can explain everything right there from the podium. With a weird project like this one, that’s important. I can give immediate feedback without a talkback mic, and stop them quickly if a note starts wrong. I dare say I'm really fast at this.
If we were recording the Hallelujah chorus, then sure, a real conductor is the way to go. But for this? Better Call Mike!
WILL IT WORK?
I sure hope so. We’re trying to do everything right - pro singers, pro studio, pro engineer, bozo conductor. I have to say, though, the cost of this is making me more nervous than I would normally be. Plus, given that “pivoting as we go” is part of the plan, as opposed to just an emergency tactic, then there’s a lot that can go wrong.
So if you see a For Sale sign on the forum next weekend, you’ll know why. Just kidding. Honestly, I’m looking forward to this. A little nervous, but I think it’s going to be fun.
A COUPLE NOTES ON HOW WE’LL RECORD
Typically, for each articulation, you start at the lowest note, then go up the scale for the full range, then move on to the next articulation. I came up with a different technique as I was recording the Realivox Ladies, though. I start with a note in the middle, then go up from there, then come back to that middle note, and then go down from there. Sort of a “Middle Out” thing. I do that because by starting in the middle, I can be sure I’m getting what I like to call “the money tone” and make adjustments if not. If you start on the lowest note, you’re in outlier territory, not really hearing what the actual articulation is going to sound like in the sweet spot until you actually get to the middle notes and realize, “Oh, this is too dark/bright/nasal/whatever!”
We did that with Sunset Strings, where it also had the advantage that since Sunset was recorded as an ensemble, it’s only the middle notes where all the cellos, violas and violins played. We’ll probably do the same thing with the choir.
I mentioned earlier that the string players could find pitch for each note without needing me to play a piano note for them. They also didn’t need to play a short tuneup to make sure their finger was in the right spot. (With no frets!)
My experience with singers is that they’re not that quick to center their pitch for each successive note. (Even more so than with instruments, recording a vocal sample library is a completely unnatural process, since singers don’t sing straight notes in real life.)
In fact, with some of the Realivox Ladies, on some of the samples, you can hear a slight ramp up at the beginnings of the notes. That’s because if they’re singing an F, then that means they had just sung an E right before that. So they still have some of that E in their head, and in the muscles of their vocal cords, so there’s that tiny little slide into the F.
So I have a new idea I might try out with the choir. With some of the basic articulations, rather than doing the articulations one at a time, where we sing a note, then sing that same articulation on the next note, I might instead focus on one note at a time. In other words, do a bunch of articulations at that one note, so their pitch will already be centered as they sing oo, then sing ah, then sing oh, all on that same note. Then go to the next note and do the same sequence. The risk of that is we may lose consistency on what exactly an oo or ah sounds like, but we’ll see.
That will only work for the basic articulations, of course, where they already know what oo and ah are, without me explaining them. For the unusual articulations, we’ll have to do those the normal way.
THE MECHANICS OF IT ALL
We make a chart (spreadsheet) of all the things we might do. We add numbers at the front of each articulation name, so they’re easy to find. For instance, we might have “01 - Oo Sustains” and then “02 - Oo Legatos.”
A single massive PT project would get unwieldy, so each of those numbers will be its own Pro Tools project. Then again, 200 separate Pro Tools projects would be a mess, too, so we’ll probably compromise with something like “01 - All Vowel Sustains” and the “02 - All Basic Legatos.”
Keeping track of what’s been recorded (and what hasn’t) gets very complicated. This is not a typical library, where you have a definitive list of what needs to get recorded. We’ll try to have a list of everything we can think of, but I guarantee we’re going to hear one of the singers doing some weird warmup that will give us ideas for more articulations.
As we add things to the list (and … cross things off! Yikes!), we’ll need to keep an eye on whether the stuff we’re keeping will make for a cohesive product. So we’ll probably be re-prioritizing what we *need* to record as we run out of time on days four and five.
Vincent will be at the sessions, and since he’s involved in the pre-production, he’ll probably be involved in managing that. Karen (our contractor) also does that sort of thing, since she’s also at all the sessions. We’ll also have Gabriel, although he’s not as familiar with the project. (He’s my nephew, and works at the studio doing other projects. Talented, but not an orchestral guy.)
Jayden is in Australia. He’ll probably be watching via Zoom, but his day is our night, so we’ll see. (Our normal Zoom meetings are at 4 in the afternoon, which is a good compromise time.) The plan is to “Audio Mover” the stereo roughs as we record, so he can make some basic patches (while we’re asleep) and let us know the next morning how things are working. That will help determine which paths we continue on, and which paths we pull the plug on.
MEET YOUR CONDUCTOR!
One of the nice things about writing the checks is you get to give all the fun jobs to yourself, and the funnest job of all is conducting. It’s stressful, but I like doing it.
Conducting the strings a couple years ago was entirely new to me, but you don’t exactly need a PhD in Conducting to move your finger in the 1-2-3-4 pattern. (I did buy a baton, but I used my finger instead.) This is mostly long sustains, so it ain’t rocket science. (Well, the full project is sorta rocket science, but the conducting isn’t.)
I’ve recorded a zillion singers over the years, and I like to believe I’m pretty good at it, partly because one of my strengths is in keeping the singer’s energy up. (With singers especially, that’s important.) There’s a whole psychology to it, where I try to stay upbeat as much as possible, because their energy feeds off of mine. (Staying upbeat for 13 hours is impossible, so I’m also hiring a real conductor to fill in for me through the day.)
Another advantage I have is that I’ve learned how to ask for re-takes in a positive way. Instead of “You were out of tune on that one. Do it again.”, I might say, “Your tone was great! Let’s do one more for tuning.” Or the old, “Let’s do one more for a safety” trick. I’ll do whatever I’ve got to do to keep them motivated and engaged. (Mind you, it’s not all trickery, and sometimes I do say simply that they were out of tune, but I’m watching for how they’re receiving that note.)
Most importantly in the “Why I conduct this myself” list, is that with this project (as with Sunset Strings), I’m the guy who knows exactly what we’re trying to accomplish. I can explain everything right there from the podium. With a weird project like this one, that’s important. I can give immediate feedback without a talkback mic, and stop them quickly if a note starts wrong. I dare say I'm really fast at this.
If we were recording the Hallelujah chorus, then sure, a real conductor is the way to go. But for this? Better Call Mike!
WILL IT WORK?
I sure hope so. We’re trying to do everything right - pro singers, pro studio, pro engineer, bozo conductor. I have to say, though, the cost of this is making me more nervous than I would normally be. Plus, given that “pivoting as we go” is part of the plan, as opposed to just an emergency tactic, then there’s a lot that can go wrong.
So if you see a For Sale sign on the forum next weekend, you’ll know why. Just kidding. Honestly, I’m looking forward to this. A little nervous, but I think it’s going to be fun.
Stringtree
Senior Member
As someone who buys these libraries, video (even haphazard amateur phone stuff) ramps things from a simmer to a boil. This is an exciting narrative. I'm interested in the process as well as the product.
If you can't release that kind of stuff video-wise to the public, at least do it for yourself; maybe someday you'll be able to. Don't make the pictures and you'll surely not have the pictures.
Your storytelling is like an old-timey serial radio show. It sets the stage and dignifies everyone involved. How absolutely cool.
If you can't release that kind of stuff video-wise to the public, at least do it for yourself; maybe someday you'll be able to. Don't make the pictures and you'll surely not have the pictures.
Your storytelling is like an old-timey serial radio show. It sets the stage and dignifies everyone involved. How absolutely cool.
ALittleNightMusic
Professional Amateur
I just saw a talk by HGW yesterday and he had an 8 vocalist choir there for a few pieces to demonstrate. I wonder if any of them will be part of your recording group. I imagine there's not a huge abundance of top tier-quality, studio-prepared singers that have experience blending with each other, even in LA.
OP
OP
WEEK #3 - THE RECORDING SESSIONS:
What kind of moron books five 13-hour days in a row? This kind of moron, that’s who! Now, I don’t take my accusations of moronhood lightly, but let’s look at the evidence:
Exhibit A - Five days is a lot to prepare for, and given that we only had two weeks to make those preparations (what was I thinking???), there was no way I would be completely ready. That turned out to be problematic in a few instances. (Which I’ll explain below.)
Exhibit B - You get tired after five days. Plus I added an extra half day on Saturday. I was pretty loopy by then. I was still gittin’ ‘er done, but not at maximum efficiency.
Now, with that said, do I regret booking a five day lockout with only two weeks to prepare? Not really. Certainly there would be some aspects of the recording sessions that might have gone more smoothly, but we got some really good stuff recorded, and we got almost everything done on the list. The sessions might not have gone quite as efficiently as I would have liked, but prep time isn’t free, either. If another week of prep time would have saved two hours of recording time … that’s not worth it to me. I’d rather pay the overage and not “waste” the week of my own time. Plus, adding weeks to the preparation period pushes back the eventual release.
This is just me, mind you, so don’t interpret this as “advice.” I have the resources to be somewhat cavalier about these things, and at this stage in my life, Realitone is more about having fun than putting my nose to the grindstone, so I like the excitement of flying by the seat of my pants.
Plus, the “Sunset” concept is so unusual that you sort of have to do a lot of this on the fly. Try something, see how it sounds, and then go from there. Too much planning can be self defeating, and a lot of it goes out the window anyway. Like giant charts that I spent hours making, but never used. What a waste.
And truth be told, a number of things I wish I were better prepared for … I’m not so sure I could have anticipated. Starting with:
VOCALS AREN’T STRINGS, FOLKS
You heard it here first! With Sunset Strings, there were names for everything: Ricochets, col legno, bartok pizz, seagulls, or whatever. You call it out, and they play it. Perfectly.
Vocals don’t have an equivalent terminology. “Sound like a weirdo” isn’t quite descriptive enough, so you have to come up with ways to explain what you’re looking for. And then (and here’s the even trickier part), make sure the men interpret your instructions the same way as the women did, so they’ll be consistent. You don't want women to have a narrow vibrato and men to have a wide one.
Dynamics are trickier with singers than with string players. For the strings, Sul Tasto was the default. For the singers, it was mp, which was a little more open to interpretation. On the bright side, this is less of an issue than it may seem, because although the singers may be slightly louder or quieter on various days, the actual timbre doesn’t change that much from p to m. (I did experiments with this ahead of time.) Still, it was something I had to keep an eye on.
The biggest difference between recording strings and vocals is that strings don’t have vowels. With Sunset Strings, we could record “Slow Bends” for attacks and releases, and they would work for most of the sustains - Sul Tasto, Normale, Flautando, even Tremolos. We did have to do some under-the-hood volume adjustments, but it all worked.
With Choir, we had to record separate attack/release versions for each vowel. So for those, we limited them to ah, oo and mm. (I think.) For the legatos, I think we went a little further and may have added uh, eh and oh. And maybe ee. (My mind is a haze right now.)
This obviously turns into a multiplication problem, where if it takes x amount of time to record something for one vowel, it take n*x amount of time to do the same for n vowels. Ugh. There are tricks, though:
EFFICIENCY, BABY!
Sampling one note at a time as you go up the scale takes time. The singer has to adjust to the new pitch, which with vocals, is a real thing. Singers have a muscle memory for the previous pitch, which often presents itself at the start of the new pitch. For normal singing, those little scoops or dips are part of the human performance, and it’s totally common, even expected. Not so much with a sample library.
Having recorded a number of singers for Realivox, I’ve handled this by having the singer sing a short centering note, and then sing the real note. It seems to work, although it takes extra time.
For this project, though, I figured why not have the singers sing multiple vowels all at the same pitch. In theory, it should be faster and have better tuning. By golly, it worked! Check it out, in this straight vowels example:
View attachment Men Sustains Day 1.mp3
When you hear me say at the end, “And now we add one tenor,” that means we moved up the scale by one note, to a B1, so the ensemble would be five basses and one tenor. That’s because although there were ten men, only five or six would ever sing at once. To explain that, let’s rewind to Day one, where we decided how to do all this:
DOWNBEAT DAY 1 - THE WOMEN
Women were in the afternoon, men in the evening. We also had morning sessions with three solo singers in the various sections, mostly for Repetitions, but that’s a story in itself. (I’ll talk about that in a future post, not here.)
The first hour of Day 1 was spent experimenting with positioning. There are 5 altos, 5 sopranos. For the low notes (we started at F2, then through A2), only the 5 altos sing. As we get higher (B2), we add one soprano, for six singers altogether. Then on the next note up (we went up by whole steps), we added another soprano and lost an alto, so we’d stay at six singers total. We did this because we didn’t want the sound to be too thick in the middle notes.
We decided to have altos right, sopranos left. (From audience perspective.) There’s some risk to this, because since we’re recording all women at once, as opposed to altos and sopranos separately, then for legato intervals, there will be a stereo shift when making a leap from a low note sung only by altos, to a higher note sung by 3 altos and 3 sopranos. The stereo imaging for that legato interval will move from right to center.
So we toyed with the idea of mixing the singers. There’s something kinda cool about separation, though, so that’s the way we went, although we spent some time positioning the women to decide how spread out they should be (seats far apart from wall to wall, or closer together, as well as interspersing them), so that the stereo imaging would be there, but not too extreme so as to be annoying with legatos. That, along with getting mic levels and stuff, took an hour or so.
That hour, plus some other preliminaries, was expected, although it meant the men now have a time advantage. The men were also a bit faster, so that’s why we added the extra half day on Saturday. (Only for the women, to catch up.)
Continued...
What kind of moron books five 13-hour days in a row? This kind of moron, that’s who! Now, I don’t take my accusations of moronhood lightly, but let’s look at the evidence:
Exhibit A - Five days is a lot to prepare for, and given that we only had two weeks to make those preparations (what was I thinking???), there was no way I would be completely ready. That turned out to be problematic in a few instances. (Which I’ll explain below.)
Exhibit B - You get tired after five days. Plus I added an extra half day on Saturday. I was pretty loopy by then. I was still gittin’ ‘er done, but not at maximum efficiency.
Now, with that said, do I regret booking a five day lockout with only two weeks to prepare? Not really. Certainly there would be some aspects of the recording sessions that might have gone more smoothly, but we got some really good stuff recorded, and we got almost everything done on the list. The sessions might not have gone quite as efficiently as I would have liked, but prep time isn’t free, either. If another week of prep time would have saved two hours of recording time … that’s not worth it to me. I’d rather pay the overage and not “waste” the week of my own time. Plus, adding weeks to the preparation period pushes back the eventual release.
This is just me, mind you, so don’t interpret this as “advice.” I have the resources to be somewhat cavalier about these things, and at this stage in my life, Realitone is more about having fun than putting my nose to the grindstone, so I like the excitement of flying by the seat of my pants.
Plus, the “Sunset” concept is so unusual that you sort of have to do a lot of this on the fly. Try something, see how it sounds, and then go from there. Too much planning can be self defeating, and a lot of it goes out the window anyway. Like giant charts that I spent hours making, but never used. What a waste.
And truth be told, a number of things I wish I were better prepared for … I’m not so sure I could have anticipated. Starting with:
VOCALS AREN’T STRINGS, FOLKS
You heard it here first! With Sunset Strings, there were names for everything: Ricochets, col legno, bartok pizz, seagulls, or whatever. You call it out, and they play it. Perfectly.
Vocals don’t have an equivalent terminology. “Sound like a weirdo” isn’t quite descriptive enough, so you have to come up with ways to explain what you’re looking for. And then (and here’s the even trickier part), make sure the men interpret your instructions the same way as the women did, so they’ll be consistent. You don't want women to have a narrow vibrato and men to have a wide one.
Dynamics are trickier with singers than with string players. For the strings, Sul Tasto was the default. For the singers, it was mp, which was a little more open to interpretation. On the bright side, this is less of an issue than it may seem, because although the singers may be slightly louder or quieter on various days, the actual timbre doesn’t change that much from p to m. (I did experiments with this ahead of time.) Still, it was something I had to keep an eye on.
The biggest difference between recording strings and vocals is that strings don’t have vowels. With Sunset Strings, we could record “Slow Bends” for attacks and releases, and they would work for most of the sustains - Sul Tasto, Normale, Flautando, even Tremolos. We did have to do some under-the-hood volume adjustments, but it all worked.
With Choir, we had to record separate attack/release versions for each vowel. So for those, we limited them to ah, oo and mm. (I think.) For the legatos, I think we went a little further and may have added uh, eh and oh. And maybe ee. (My mind is a haze right now.)
This obviously turns into a multiplication problem, where if it takes x amount of time to record something for one vowel, it take n*x amount of time to do the same for n vowels. Ugh. There are tricks, though:
EFFICIENCY, BABY!
Sampling one note at a time as you go up the scale takes time. The singer has to adjust to the new pitch, which with vocals, is a real thing. Singers have a muscle memory for the previous pitch, which often presents itself at the start of the new pitch. For normal singing, those little scoops or dips are part of the human performance, and it’s totally common, even expected. Not so much with a sample library.
Having recorded a number of singers for Realivox, I’ve handled this by having the singer sing a short centering note, and then sing the real note. It seems to work, although it takes extra time.
For this project, though, I figured why not have the singers sing multiple vowels all at the same pitch. In theory, it should be faster and have better tuning. By golly, it worked! Check it out, in this straight vowels example:
View attachment Men Sustains Day 1.mp3
When you hear me say at the end, “And now we add one tenor,” that means we moved up the scale by one note, to a B1, so the ensemble would be five basses and one tenor. That’s because although there were ten men, only five or six would ever sing at once. To explain that, let’s rewind to Day one, where we decided how to do all this:
DOWNBEAT DAY 1 - THE WOMEN
Women were in the afternoon, men in the evening. We also had morning sessions with three solo singers in the various sections, mostly for Repetitions, but that’s a story in itself. (I’ll talk about that in a future post, not here.)
The first hour of Day 1 was spent experimenting with positioning. There are 5 altos, 5 sopranos. For the low notes (we started at F2, then through A2), only the 5 altos sing. As we get higher (B2), we add one soprano, for six singers altogether. Then on the next note up (we went up by whole steps), we added another soprano and lost an alto, so we’d stay at six singers total. We did this because we didn’t want the sound to be too thick in the middle notes.
We decided to have altos right, sopranos left. (From audience perspective.) There’s some risk to this, because since we’re recording all women at once, as opposed to altos and sopranos separately, then for legato intervals, there will be a stereo shift when making a leap from a low note sung only by altos, to a higher note sung by 3 altos and 3 sopranos. The stereo imaging for that legato interval will move from right to center.
So we toyed with the idea of mixing the singers. There’s something kinda cool about separation, though, so that’s the way we went, although we spent some time positioning the women to decide how spread out they should be (seats far apart from wall to wall, or closer together, as well as interspersing them), so that the stereo imaging would be there, but not too extreme so as to be annoying with legatos. That, along with getting mic levels and stuff, took an hour or so.
That hour, plus some other preliminaries, was expected, although it meant the men now have a time advantage. The men were also a bit faster, so that’s why we added the extra half day on Saturday. (Only for the women, to catch up.)
Continued...
OP
OP
continued ...
WAIT … THE MEN WERE A BIT FASTER? WHY?
The women had a slightly larger range, requiring an additional note for each articulation. (Basses and altos both started on an F, but the sopranos went up to an A, while the tenors only went up to a G.)
Not only do the women sing an additional note, but that high A is harder to sing than any of the other notes, so it took more re-takes than the other notes. Even the high G was more challenging than lower notes, bearing in mind that sample sessions are a very unnatural way of singing, so sustaining an “Ee,” and with me expecting a perfect entrance from all five singers … that’s harder than you might think, even for professionals.
Tenors don't have these issues with their high notes, since their high G is actually nowhere near the end of their range.
This time differential turned out to be significant enough that I added an extra half day on Saturday so the women could catch up. We had to do it. (It’s only money, right?)
SO HOW DID THEY SOUND???
I promised when I started this thread that I wouldn’t gush, so I won’t. But … let’s just say I was really happy. We’re going for a tone between classical and pop. I didn’t want a diva vibe, or those exaggerated bass tones, but I also wanted it to sound like an actual choir, not the background singers for a Partridge Family record. The goal was to sound in the ballpark of Eric Whitacre’s choir.
Our photographer wanted to snap some pictures with all ten singers singing. As I explained earlier, this is a 6-singer-max situation, so he wasn't getting any "all singers" shots. So we decided, just for the sake of pictures, to have them sing Amazing Grace. Useless for the project, but good for the photographer, plus it gives a good mix reference for later.
We printed up lyric sheets, but there was no rehearsal for this. I counted them off and away they went. They figured out harmonies on the fly (although undoubtedly they’ve sung this song before) and it was really interesting to watch them signal to each other as they adjusted who should sing what. (You'll notice the first verse harmonies are not the same as the last verse.)
It was incredibly cool to have a front row seat for this, as it felt like I was being serenaded. Check it out:
View attachment Amazing Grace.mp3
SLOW BENDS, ANYONE?
The slow bends were a hook with Sunset Strings, so everyone will be expecting them here. We went for more intervals, and did them for oo, ah and mm. Here’s the men doing a “2-beat” bend to the 3rd:
View attachment Men Slow Bends Third.mp3
These are much more difficult for singers than for instrumentalists, since singers don’t have a fingerboard to look at to make sure they’re timing their rise to “land” at the right time. Challenges like these are why this took five and a half days.
The slow bends were one of the few times we used headphones with a click from the booth, because the timings were so difficult otherwise. You generally don’t want singers wearing headphones, because they need to be listening to each other as much as possible. (Even with only one earpiece on, they’re missing one side.)
THE MIC POSITIONINGS
You can see the “close mics” (including the stereo overheads) in these pictures, although they’re obviously not that close. I can’t remember what Damon (Tedesco, our engineer) used for those. (He brings his own mics for that.) In the empty room picture, you can see two of the room mics (left room and center room) which are Telefunken 251s in omni position. (In the audio examples, you'll notice that my voice is very loud. That's because I'm very close to the center room mic.)
Those same mics are also at the back of the room, which are our ambient mics. This is a near identical setup to Sunset Strings.
You’ll also note that the singers are in two rows, men in the back. Since the men and women were never there at the same time, they could have shared a single row, but we thought (or rather, Damon thought) the sound of the men being slightly back would be good.
ARE YOU AVOIDING SOMETHING, MIKE? DIDN’T YOU SAY SOMETHING ABOUT MISTAKES YOU MADE?
Yep, there were plenty. Mainly that I wasn’t as prepared as I wish I had been, mostly in terms of knowing what articulations I wanted to record. I probably focused more than I should have on the word-builder aspects. (Which was a total success, by the way. After Sunset Choirs, we’ll be moving forward with that.)
I was constantly behind on setting the days' agendas, so for a couple hours each morning, and during every break, I was looking at the lists of things we wanted to record, and not only trying to organize them, but in many cases, I was trying to understand what some of them even were. As I mentioned earlier, you can’t simply write “ricochets.” Conceptually, this was a very difficult library to define, and descriptions that made sense in the two weeks of preparation were much hazier now.
Then there was the issue of men and women recording separately. When we recorded an ensemble (women, for example) where various singers would intentionally drift out of tune for a haunting effect, then when the men came in to do the same thing, we’d want to make sure their out-of-tuneness was at a similar level, so it would match. That’s tricky.
In hindsight, I should have recorded a few notes of each of these myself on the fly, so I could play them back for the other ensemble to reference.
We could have accomplished the same thing by pulling references from the Pro Tools sessions, of course, and in fact, we did that a few times. But it’s tricky, because I’d have to communicate that to Gabriel (another assistant - also my nephew), and given that Gabriel is less savvy about this project than Vincent or Jayden, explaining exactly what I needed was difficult. It’s all possible, mind you, but this was a situation where every minute counts, so if it took me five minutes to explain what I wanted edited before the evening session, there’s a cost to that, in terms of what other things I wouldn’t have time to do during “breaks.” I was often overwhelmed by all the things I was trying to juggle, which cost me.
WE CERTAINLY COULDN’T AFFORD TO WASTE TIME!
The women were recording some short “words,” which we did in an alphabetical sequence per vowel. For short a, they might sing: bab, chach, cack, dad, faff, gag, etc. Simple sequence by consonant, right? But for oo, if we continue the same start/end letter trend, the first word would be “boob.” I’d rather avoid that (along with tit, cock, poop, etc.), so instead of boob and chooch, I reversed them and they sang booch and choob.
After the sequence, Damon got on the talkback mic and asked if we could do a single “chooch,” which apparently is Italian slang for … well, let's just say it’s not a good thing to be a chooch. He wanted a recording to use as a ringtone for when certain people call him. Easy enough, so we did it and he got a kick out of it.
Everybody likes Damon, so we decided to take it a step further and give him an even better ringtone. Vincent secretly wrote up a little choir arrangement for "This Guy is Such a Chooch," and on Friday, the women sang it in the afternoon, and the men sang their parts that evening. It was all a secret that Damon didn’t realize was happening until the women actually sang it. Check out the mix:
View attachment This Guy Is a Chooch.mp3
SO ... WAS RECORDING A CHOIR WORTH IT?
Definitely. The main thing I was nervous about was wondering how they would sound, and would they be reasonably fast. These are pros, but still, you worry, because if there's a problem, I'm locked into five days of that problem. Yikes!
Pretty early on, though, I knew this would go well. Really well, in fact. This is a great group of singers, I love their tone, and we got more recorded than I expected. Almost everything on the list, in fact.
But ... almost everything isn't the same as everything everything. I'm greedy, so I want everything everything. Plus I got more ideas since the recordings, and I'm sure Jayden and Vincent have, too. (This stuff gets you excited!) We definitely need to do some pickup sessions anyway for the smaller breakout solo groups. (Three hours each morning simply wasn't enough.) That will probably be in a couple months. So I'm kinda tempted to add on the main ensembles as well. We'll see ...
WAIT … THE MEN WERE A BIT FASTER? WHY?
The women had a slightly larger range, requiring an additional note for each articulation. (Basses and altos both started on an F, but the sopranos went up to an A, while the tenors only went up to a G.)
Not only do the women sing an additional note, but that high A is harder to sing than any of the other notes, so it took more re-takes than the other notes. Even the high G was more challenging than lower notes, bearing in mind that sample sessions are a very unnatural way of singing, so sustaining an “Ee,” and with me expecting a perfect entrance from all five singers … that’s harder than you might think, even for professionals.
Tenors don't have these issues with their high notes, since their high G is actually nowhere near the end of their range.
This time differential turned out to be significant enough that I added an extra half day on Saturday so the women could catch up. We had to do it. (It’s only money, right?)
SO HOW DID THEY SOUND???
I promised when I started this thread that I wouldn’t gush, so I won’t. But … let’s just say I was really happy. We’re going for a tone between classical and pop. I didn’t want a diva vibe, or those exaggerated bass tones, but I also wanted it to sound like an actual choir, not the background singers for a Partridge Family record. The goal was to sound in the ballpark of Eric Whitacre’s choir.
Our photographer wanted to snap some pictures with all ten singers singing. As I explained earlier, this is a 6-singer-max situation, so he wasn't getting any "all singers" shots. So we decided, just for the sake of pictures, to have them sing Amazing Grace. Useless for the project, but good for the photographer, plus it gives a good mix reference for later.
We printed up lyric sheets, but there was no rehearsal for this. I counted them off and away they went. They figured out harmonies on the fly (although undoubtedly they’ve sung this song before) and it was really interesting to watch them signal to each other as they adjusted who should sing what. (You'll notice the first verse harmonies are not the same as the last verse.)
It was incredibly cool to have a front row seat for this, as it felt like I was being serenaded. Check it out:
View attachment Amazing Grace.mp3
SLOW BENDS, ANYONE?
The slow bends were a hook with Sunset Strings, so everyone will be expecting them here. We went for more intervals, and did them for oo, ah and mm. Here’s the men doing a “2-beat” bend to the 3rd:
View attachment Men Slow Bends Third.mp3
These are much more difficult for singers than for instrumentalists, since singers don’t have a fingerboard to look at to make sure they’re timing their rise to “land” at the right time. Challenges like these are why this took five and a half days.
The slow bends were one of the few times we used headphones with a click from the booth, because the timings were so difficult otherwise. You generally don’t want singers wearing headphones, because they need to be listening to each other as much as possible. (Even with only one earpiece on, they’re missing one side.)
THE MIC POSITIONINGS
You can see the “close mics” (including the stereo overheads) in these pictures, although they’re obviously not that close. I can’t remember what Damon (Tedesco, our engineer) used for those. (He brings his own mics for that.) In the empty room picture, you can see two of the room mics (left room and center room) which are Telefunken 251s in omni position. (In the audio examples, you'll notice that my voice is very loud. That's because I'm very close to the center room mic.)
Those same mics are also at the back of the room, which are our ambient mics. This is a near identical setup to Sunset Strings.
You’ll also note that the singers are in two rows, men in the back. Since the men and women were never there at the same time, they could have shared a single row, but we thought (or rather, Damon thought) the sound of the men being slightly back would be good.
ARE YOU AVOIDING SOMETHING, MIKE? DIDN’T YOU SAY SOMETHING ABOUT MISTAKES YOU MADE?
Yep, there were plenty. Mainly that I wasn’t as prepared as I wish I had been, mostly in terms of knowing what articulations I wanted to record. I probably focused more than I should have on the word-builder aspects. (Which was a total success, by the way. After Sunset Choirs, we’ll be moving forward with that.)
I was constantly behind on setting the days' agendas, so for a couple hours each morning, and during every break, I was looking at the lists of things we wanted to record, and not only trying to organize them, but in many cases, I was trying to understand what some of them even were. As I mentioned earlier, you can’t simply write “ricochets.” Conceptually, this was a very difficult library to define, and descriptions that made sense in the two weeks of preparation were much hazier now.
Then there was the issue of men and women recording separately. When we recorded an ensemble (women, for example) where various singers would intentionally drift out of tune for a haunting effect, then when the men came in to do the same thing, we’d want to make sure their out-of-tuneness was at a similar level, so it would match. That’s tricky.
In hindsight, I should have recorded a few notes of each of these myself on the fly, so I could play them back for the other ensemble to reference.
We could have accomplished the same thing by pulling references from the Pro Tools sessions, of course, and in fact, we did that a few times. But it’s tricky, because I’d have to communicate that to Gabriel (another assistant - also my nephew), and given that Gabriel is less savvy about this project than Vincent or Jayden, explaining exactly what I needed was difficult. It’s all possible, mind you, but this was a situation where every minute counts, so if it took me five minutes to explain what I wanted edited before the evening session, there’s a cost to that, in terms of what other things I wouldn’t have time to do during “breaks.” I was often overwhelmed by all the things I was trying to juggle, which cost me.
WE CERTAINLY COULDN’T AFFORD TO WASTE TIME!
The women were recording some short “words,” which we did in an alphabetical sequence per vowel. For short a, they might sing: bab, chach, cack, dad, faff, gag, etc. Simple sequence by consonant, right? But for oo, if we continue the same start/end letter trend, the first word would be “boob.” I’d rather avoid that (along with tit, cock, poop, etc.), so instead of boob and chooch, I reversed them and they sang booch and choob.
After the sequence, Damon got on the talkback mic and asked if we could do a single “chooch,” which apparently is Italian slang for … well, let's just say it’s not a good thing to be a chooch. He wanted a recording to use as a ringtone for when certain people call him. Easy enough, so we did it and he got a kick out of it.
Everybody likes Damon, so we decided to take it a step further and give him an even better ringtone. Vincent secretly wrote up a little choir arrangement for "This Guy is Such a Chooch," and on Friday, the women sang it in the afternoon, and the men sang their parts that evening. It was all a secret that Damon didn’t realize was happening until the women actually sang it. Check out the mix:
View attachment This Guy Is a Chooch.mp3
SO ... WAS RECORDING A CHOIR WORTH IT?
Definitely. The main thing I was nervous about was wondering how they would sound, and would they be reasonably fast. These are pros, but still, you worry, because if there's a problem, I'm locked into five days of that problem. Yikes!
Pretty early on, though, I knew this would go well. Really well, in fact. This is a great group of singers, I love their tone, and we got more recorded than I expected. Almost everything on the list, in fact.
But ... almost everything isn't the same as everything everything. I'm greedy, so I want everything everything. Plus I got more ideas since the recordings, and I'm sure Jayden and Vincent have, too. (This stuff gets you excited!) We definitely need to do some pickup sessions anyway for the smaller breakout solo groups. (Three hours each morning simply wasn't enough.) That will probably be in a couple months. So I'm kinda tempted to add on the main ensembles as well. We'll see ...
Reznov981
Senior Member
LOVE these updates Mike, for real!
Just wanted to suggest maybe a feature button/switch to stretch the range up and down a tone or two, or something. I've seen it on a couple vocal libraries. It doesn't sound amazing but it can get you out of a jam if your idea just wants to go up that extra step beyond the normal range. Totally up to you of course.
Edit: now read week 3. Again, such a joy to get a look behind the scenes like this. No other developer is this generous and transparent. It goes a long way! I'm loving this a lot. Thank you!
Last edited:
OP
OP
Thanks for the responses! I couldn't really respond until now, because time was so tight. I'm in recuperate mode now, so:
The general planning was that the most important articulations were done in the first days, since if someone gets sick or something halfway through (fortunately no one did), we'd have the most important things already recorded.
Those "most important" things are the ones where it would be most noticeable if there was a change in singers. Oo and ah, for instance. If oo is on Layer 1 and ah is on Layer 2, we want the singers to be consistent as the mod wheel moves from one to the other. Same with the attack and release articulations. The voices need to match. Whereas with a more bizarre articulation, that would be less noticeable, so we leaned towards having those at the end of the week.
Yep. David Kudell came in on Wednesday and got some great footage. It's a lot to look through, and I'm not sure what we'll actually do with it, but I'm glad we did that.
Not very accurately. Some things are quicker than we expect and others take longer. Although these weren't union sessions, we abided by union rules, including 10 minute breaks each hour. So if we finished an articulation and had ten minutes left in the hour, I might move to a really quick articulation that might not have been on the day's agenda. Or we might start a regular articulation and just finish the first few notes before taking a break.
The general planning was that the most important articulations were done in the first days, since if someone gets sick or something halfway through (fortunately no one did), we'd have the most important things already recorded.
Those "most important" things are the ones where it would be most noticeable if there was a change in singers. Oo and ah, for instance. If oo is on Layer 1 and ah is on Layer 2, we want the singers to be consistent as the mod wheel moves from one to the other. Same with the attack and release articulations. The voices need to match. Whereas with a more bizarre articulation, that would be less noticeable, so we leaned towards having those at the end of the week.
Not trying to be a salesman, but this won't be a typical oos and ahs library.
That's the fear, because none of my costs are refundable, and especially in times of covid, things can go terribly wrong.
Good idea. I've now added them to the first post.
I tend to think the same way, and that's what we did with the Realivox libraries, which have extendable ranges. We'll mostly likely do the same with Sunset Choirs.
moon
Active Member
I'm loving these write ups, Mike! I'm curious how you decided the order of the vowels to sing. It seems reasonable to me that changing the order would change the performance - for example, singing 'oh' when preceded by 'ooh' will sound different from if it is preceded by 'ah'. Obviously, this might not make a lick of a difference, but I'm curious if it's something that matters at all, and if you might have experimented with it.WEEK #3 - THE RECORDING SESSIONS:
What kind of moron books five 13-hour days in a row? This kind of moron, that’s who! Now, I don’t take my accusations of moronhood lightly, but let’s look at the evidence:
Exhibit A - Five days is a lot to prepare for, and given that we only had two weeks to make those preparations (what was I thinking???), there was no way I would be completely ready. That turned out to be problematic in a few instances. (Which I’ll explain below.)
Exhibit B - You get tired after five days. Plus I added an extra half day on Saturday. I was pretty loopy by then. I was still gittin’ ‘er done, but not at maximum efficiency.
Now, with that said, do I regret booking a five day lockout with only two weeks to prepare? Not really. Certainly there would be some aspects of the recording sessions that might have gone more smoothly, but we got some really good stuff recorded, and we got almost everything done on the list. The sessions might not have gone quite as efficiently as I would have liked, but prep time isn’t free, either. If another week of prep time would have saved two hours of recording time … that’s not worth it to me. I’d rather pay the overage and not “waste” the week of my own time. Plus, adding weeks to the preparation period pushes back the eventual release.
This is just me, mind you, so don’t interpret this as “advice.” I have the resources to be somewhat cavalier about these things, and at this stage in my life, Realitone is more about having fun than putting my nose to the grindstone, so I like the excitement of flying by the seat of my pants.
Plus, the “Sunset” concept is so unusual that you sort of have to do a lot of this on the fly. Try something, see how it sounds, and then go from there. Too much planning can be self defeating, and a lot of it goes out the window anyway. Like giant charts that I spent hours making, but never used. What a waste.
And truth be told, a number of things I wish I were better prepared for … I’m not so sure I could have anticipated. Starting with:
VOCALS AREN’T STRINGS, FOLKS
You heard it here first! With Sunset Strings, there were names for everything: Ricochets, col legno, bartok pizz, seagulls, or whatever. You call it out, and they play it. Perfectly.
Vocals don’t have an equivalent terminology. “Sound like a weirdo” isn’t quite descriptive enough, so you have to come up with ways to explain what you’re looking for. And then (and here’s the even trickier part), make sure the men interpret your instructions the same way as the women did, so they’ll be consistent. You don't women to have a narrow vibrato and men to have a wide one.
Dynamics are trickier with singers than with string players. For the strings, Sul Tasto was the default. For the singers, it was mp, which was a little more open to interpretation. On the bright side, this is less of an issue than it may seem, because although the singers may be slightly louder or quieter on various days, the actual timbre doesn’t change that much from p to m. (I did experiments with this ahead of time.) Still, it was something I had to keep an eye on.
The biggest difference between recording strings and vocals is that strings don’t have vowels. With Sunset Strings, we could record “Slow Bends” for attacks and releases, and they would work for most of the sustains - Sul Tasto, Normale, Flautando, even Tremolos. We did have to do some under-the-hood volume adjustments, but it all worked.
With Choir, we had to record separate attack/release versions for each vowel. So for those, we limited them to ah, oo and mm. (I think.) For the legatos, I think we went a little further and may have added uh, eh and oh. And maybe ee. (My mind is a haze right now.)
This obviously turns into a multiplication problem, where if it takes x amount of time to record something for one vowel, it take n*x amount of time to do the same for n vowels. Ugh. There are tricks, though:
EFFICIENCY, BABY!
Sampling one note at a time as you go up the scale takes time. The singer has to adjust to the new pitch, which with vocals, is a real thing. Singers have a muscle memory for the previous pitch, which often presents itself at the start of the new pitch. For normal singing, those little scoops or dips are part of the human performance, and it’s totally common, even expected. Not so much with a sample library.
Having recorded a number of singers for Realivox, I’ve handled this by having the singer sing a short centering note, and then sing the real note. It seems to work, although it takes extra time.
For this project, though, I figured why not have the singers sing multiple vowels all at the same pitch. In theory, it should be faster and have better tuning. By golly, it worked! Check it out, in this straight vowels example:
View attachment Men Sustains Day 1.mp3
When you hear me say at the end, “And now we add one tenor,” that means we moved up the scale by one note, to a B1, so the ensemble would be five basses and one tenor. That’s because although there were ten men, only five or six would ever sing at once. To explain that, let’s rewind to Day one, where we decided how to do all this:
DOWNBEAT DAY 1 - THE WOMEN
Women were in the afternoon, men in the evening. We also had morning sessions with three solo singers in the various sections, mostly for Repetitions, but that’s a story in itself. (I’ll talk about that in a future post, not here.)
The first hour of Day 1 was spent experimenting with positioning. There are 5 altos, 5 sopranos. For the low notes (we started at F2, then through A2), only the 5 altos sing. As we get higher (B2), we add one soprano, for six singers altogether. Then on the next note up (we went up by whole steps), we added another soprano and lost an alto, so we’d stay at six singers total. We did this because we didn’t want the sound to be too thick in the middle notes.
We decided to have altos right, sopranos left. (From audience perspective.) There’s some risk to this, because since we’re recording all women at once, as opposed to altos and sopranos separately, then for legato intervals, there will be a stereo shift when making a leap from a low note sung only by altos, to a higher note sung by 3 altos and 3 sopranos. The stereo imaging for that legato interval will move from right to center.
So we toyed with the idea of mixing the singers. There’s something kinda cool about separation, though, so that’s the way we went, although we spent some time positioning the women to decide how spread out they should be (seats far apart from wall to wall, or closer together, as well as interspersing them), so that the stereo imaging would be there, but not too extreme so as to be annoying with legatos. That, along with getting mic levels and stuff, took an hour or so.
That hour, plus some other preliminaries, was expected, although it meant the men now have a time advantage. The men were also a bit faster, so that’s why we added the extra half day on Saturday. (Only for the women, to catch up.)
Continued...
Reid Rosefelt
aka Tiger the Frog
The writing you are doing on this is so good and so rare. It's funny and moving (Amazing Grace) too and it shows your personality. We don't have this kind of vivid reportage on any sample library I can think of. So if you combine the kind of stuff you've written with the footage, you could have a very special and memorable video. Or videos. It will be shared a lot when the library comes out.
Once this stuff is up on YouTube, it will stay there as long as there is a YouTube. It will turn up every time somebody does a Google search for the library. And I think you'll get sales out of it for years and years.
Obviously it will be terrific on your site too.
If you don't have the time or inclination to edit it yourself, there are a lot of freelance editors available online, and I bet quite a few right here on VI:Control.
ALittleNightMusic
Professional Amateur
Btw, I just came across a series of interviews David Das did with Eric Whitacre on the creation of his choir with Spitfire. Thought it might be interesting and relevant to this thread:
