Insights In Sound 205 – Nate Baglyos, Kali Audio – S21 E5

Welcome to Insights and Sound, where we talk to the people behind the scenes, behind the technology, and behind the
0:06
music. People you may not know, but you should. And please check out
0:11
insightsandsound.com for a full list of our podcasts and YouTube series. Insights and Sound is made possible with
0:18
the support of our friends GC Customhouse.
0:25
I’m here with Nate Baglios at Cali Audio and we want to talk about voicing speakers. Yeah.
0:31
Which is something that is not necessarily unique to you guys, but you have a unique way of approaching it, I
0:37
would say. Yeah. I I would hesitate to say unique because there are people who are using
0:42
similar design philosophies. Um I I don’t know that we’re the only people uh doing things the way that we’re doing
0:48
them, but we do have a process that we feel really good about. Um, and we have science behind our design philosophies
0:55
which would support that if we’re doing what we say we’re doing correctly, you should have a really good translation
1:01
from our speakers. Science. You guys use science. Yeah, a little bit.
1:07
Oh man, it takes all the fun out of it. Takes all the mystery out of it’s supposed to be black magic. No, there’s still plenty of mystery and
1:13
there’s still plenty of uh art and finesse in all this. Um but there is
1:18
there is a goal in terms of how should the speaker sound in a room um that goal
1:24
is largely agreed upon um by most people in the industry. The question is how do you achieve that goal right
1:30
and I think we’ve got some ways that we go about it um that are tremendously effective.
1:36
Do tell. If you talk to our engineers Charles in particular one of the things he’ll talk your ears
1:41
buckle up first. Yeah. Yeah. ears off about is uh starting sort
1:46
of what an ideal speaker looks like from design principles versus starting it from materials and components. Um so
1:53
there are some things that we can do that don’t cost any additional money which greatly enhance the performance of
1:58
the loudspeaker. The easiest things to point to would be the wave guide that is around the tweeter or in the case of our
2:04
three-way speakers around the mid-range and tweeter section and the port tube that’s in the front of the speaker. The
2:09
wave guide specifically has to do with how the speaker plays in space. So the idea behind the wave guide that we have
2:15
designed is that the woofer has a certain directivity characteristic and as we get into the crossover point that
2:22
directivity uh is starting to pull inward because higher frequencies have much greater directivity in general. Right.
2:29
In general. Yeah. In general. Um blanket statement. Blanket statement. grossly oversimplifying, but yes, in general,
2:35
they have they have uh they are more directive than lower frequencies. So, as we move up to the crossover point,
2:42
particularly on a two-way speaker, the woofer is really starting to pull in in terms of its directivity. The tweeter,
2:48
by contrast, uh is omnidirectional at the crossover point. Um it’s got it’s
2:54
doing like a 1500 Hz wave compared to the size of the speaker. That wave is far larger. So, it’s projecting in 360°.
3:01
So now you have this mismatch between what the woofer is doing in space and what the tweeter is doing in space.
3:06
Ah okay. You can hear that mismatch. And even if you’re thinking, “Oh, I don’t have the
3:12
best ears in the world. I’m sure you know uh the great I’m sure Andrew Sheps
3:18
can hear that. I’m sure um uh Matthew Wallace can hear that. But I don’t
3:23
that’s not me.” No, everybody can hear that. It’s a perception thing. Even if you don’t know what you’re hearing,
3:28
even if you don’t know what you’re hearing, you are hearing it. And how it comes across is your brain is
3:34
extraordinarily well attuned to separating signal sound from noise
3:39
sound. Um, so if you and I are having a conversation, right now we’re in a fairly quiet room, it’s easy to for my
3:45
brain to tell I’m listening to you. But even if we were in a larger room, even if we were surrounded by people who are
3:51
also having conversations, my brain is really good at just listening to the words coming out of your mouth. The
3:56
similar thing happens when you’re listening to a speaker. um when you hear the speaker in the room, your brain
4:02
knows that you’re listening to the speaker. Where this breaks down, okay, is if you’re listening to a mix, if
4:08
you’re listening to a stereo mix or any sort of immersive format, there are things that are not just assigned to the
4:14
left channel and the right channel. There’s almost always a phantom center. So, anything that is that is going to
4:20
both speakers in equal amounts should render as center. Um, and a really good
4:25
mix has depth to it. It has height to it. It has width that probably extends beyond the speakers themselves. And this
4:32
is important. This is not just an issue of, oh, for people mixing in the best mixing environments so that people are
4:39
listening in the best hi-fi environments, they can enjoy this. This trickles down to every mix that is ever
4:46
produced because the person who is mixing has to be able to tell how all the elements of their mix are
4:51
interacting. Absolutely. And if they’re doing so in sort of this squished space that is just these two speakers, they’re doing so
4:58
without all the tools that they need. It would be sort of trying to paint the side of a building uh looking through
5:03
lenses that were way too small. Good analogy. Yeah. Um the listening environment is not
5:08
always ideal, but more and more listening environments are getting better and better. Certainly, anybody listening on headphones is almost always
5:15
hearing a very good stereo field. Um, anybody listening on a nice soundbar at home, um, or a nice hi-fi system at home
5:22
is also getting a very good stereo field and they should be able to interpret all the elements in the mix the way that it
5:28
was intended from the mix engineer. And if the mix engineer doesn’t have the tools to do that, well, the mix isn’t
5:34
going to sound good on those on those better systems. And I would argue at this point, everybody’s got a decent
5:39
pair of headphones. Even if we don’t, we have mixes that are designed for those earbuds or whatever
5:46
else we’re wearing. Yeah. My point in all this is uh if the tweeter and the woofer don’t match each other, your brain knows immediately
5:53
that’s coming from the speakers. If the tweeter and the woofer do match each other, that mimics a much more natural
5:58
sound source. Most sound sources in nature do not have weird phase cancellations between high frequency and
6:04
low frequency. Um, and uh, your brain starts to build that all this integrated
6:11
information that is bouncing off the walls, bouncing off the ceiling and the floor and other objects in your room is
6:16
part of the signal. It’s not part of noise. Um, so that’s when it builds that
6:22
that stereo image, that phantom center that really starts to make it easy to pick apart where are my elements in my
6:29
mix? Where were microphones placed in the room? How big was the room? Does
6:35
this recording that the guitarist did in a tiny little closet match this recording that the drummer did in a
6:40
church? Absolutely, it doesn’t. And I can hear that and I can make adjustments in the mix based on that to make it
6:47
sound like they’re playing together. Okay. Now, you mentioned something a moment ago about height, which I think
6:55
is much more applicable in a multi- channelannel like an Atmos environment than it is in a stereo environment.
7:02
Yeah. Typically, the thing about stereo is, you know, we are binaural beings. We
7:08
are stereo beings at the end of the day. Um, we have two receivers to tell us
7:13
where the sound source is coming from. So if you are in a room with speakers that have excellent directivity um where
7:21
the room is not fighting that directivity, your brain will render three dimensions,
7:26
right? Of course that’s a big if, but we won’t go into that. It’s a big if. Um but it is possible. Sure.
7:31
And you can hear this on really good recordings. Can you really discern? I mean because that you know that’s one of the things
7:36
that always struck me about immersive mixes and especially Atmos but
7:42
also I mean you know any of the other immersive formats as well um the fact
7:47
that it gives you a certain much more of a depth of height or a dimension of
7:53
height rather that you did not not necessarily have as much of with I mean
7:59
stereo is great for pan and depth. Yeah. you know, but not as much for
8:04
height. I think I’ve experienced the height. Have you? I have. On two speakers. Yes. Interesting. Okay. Um, yeah, I’ve absolutely experienced
8:11
the height. I can’t speak to, you know, what is in the signal that is causing that? What are our ears reproducing that
8:17
is allowing that to happen, but I have experienced the height. What I will say is that when you’re working in an immersive format, you’re
8:22
not nearly as reliant on the speakers working perfectly with the room and the
8:27
mix still including that information. Sure. because then you’ve got objects and you’re placing them wherever they
8:32
want to be in that field and it’s a it’s an immersive field. Yeah. But yeah, it’s interesting that um I’ve
8:39
you know because that’s something that I used to struggle with in a mix was you know in mixing in stereo u obtaining
8:46
that sort of height perspective you know because that’s the only other way we have to separate a mix. Sure. You know,
8:52
and that’s one of the one of the challenging things about doing a really good mix is, you know, how do you get
8:58
everything to to separate and stand out? Assuming that’s what you want, right? I mean, there are some people who want it to be mushed together in certain cases,
9:05
but if you really want to be able to quote unquote hear everything, how do you get that separation? You’ve
9:11
got to go for either, you know, band, you know, EQ or separation in terms of
9:16
width or depth. It’s pretty much all you got. Here’s what I’ll say. Um because this was the clearest experience I had of it
9:23
listening to classical recordings that were done with a decat tree. Okay. Yeah. Where and maybe this is a little bit of
9:30
conditioning as well cuz I know where everybody is sitting. Um but the
9:35
orchestra was so clearly rendered in three dimensions. Um and these are recordings from like the 1950s.
9:41
Oh, sure. Yeah. Um the great Al Schmidtz kind of stuff. Yeah. Mhm. Um, and it works and I’ve heard it and I
9:47
know stereo does have that promise, but it’s hard. It’s hard. I mean, you know, there’s like midside is a really great way of
9:54
achieving some of that stuff. Absolutely. And uh, you know, there’s um
10:00
there’s the technique of doing it with one mic where you’ve got one stereo mic and you do back and forth, you know,
10:05
back uh front and back basically and stuff like that. But it is a lot harder.
10:10
Definitely. Yeah. Mhm. Yeah. Yeah. I mean, we can just throw more channels at the problem and suddenly we’ve got all this bandwidth to
10:16
play with, which is cool. I like that. Um, but it’s not necessary.
10:21
Okay. So, how do you go about obtaining that objectivity in the design
10:28
of the speaker? Yeah. So, a lot of that is based on um research that is published um you know,
10:34
staying current on AAS white papers um and then also interfacing with the people who are using our products. Um so
10:41
I’ll give an example. Um how Charles’s design philosophy has changed over time.
10:46
Um he went from expecting an in room response uh with a knee in it. So around
10:52
a thousand hertz um there would be a knee and below that knee in room response would be down uh a dB or two at
11:00
the top of the human frequency range. That’s a pretty typical design. Pretty typical design. Yeah. Um what
11:08
he eventually found through some extra research that was done is that even if there is not necessarily more energy on
11:16
either side of that knee, we perceive it as such. Um because there’s an inflection point.
11:22
Okay. Um so we hear that knee as you know if it’s a positive knee we hear it as a honk. If it’s a negative knee we hear it
11:29
as a scoop even though that’s not necessarily the case. Okay. um so looking into what is going
11:35
on psycho acoustically and we’ve um our design philosophy reflects that change in his own understanding of how do we
11:42
perceive sound from a speaker in a room. Um one of the other things that we’ve come into particularly when we’re doing
11:48
room tuning is having an open mind in terms of what the slope ought to be in
11:54
terms of sound power from 20 hertz to 20 kHz which of course is going to differ with the room too
11:59
which is going to differ with the room. Um, and what really good room tuners will tell you is don’t fight the room.
12:05
Oh, absolutely. Um, absolutely. There is no one target. Every speaker in
12:10
every room should sound this way. You look at the way the speaker is already fight sounding in the room and you apply
12:16
minimal correction to get it to line up to that nice smooth uh uh slope.
12:22
Right. And it and you know, I mean, it’s uh it’s interesting. I came up in uh Cherokee where we had four different
12:29
rooms and they were very very different, you know. I mean, the mix room had the old-fashioned typical rock wall in the
12:36
front, live and deadend kind of thing, you know. There was one room that was much more of a tracking room, but it was
12:41
all wood, you know, and completely different dimensions. Of course, none of them I mean, all of them asymmetrical,
12:47
but in very, very different ways. And and I know that, you know, we had a guy that used to come around and tune the
12:52
room, uh, Coco. Okay. And uh and you know, he would always
12:58
talk about the fact that every single one of these rooms was so completely different that you know, he really had
13:04
to approach each one very differently in terms of how he tuned them. Yeah. You know, yeah,
13:10
you can’t get rid of the room. You can’t No, you can’t get rid of the room. And we are in, you know, we are mixing in rooms. We’re not mixing in an
13:17
anoic environment. And we assume that people aren’t listening in an anooic environment. Um, so yeah, we are trying
13:23
to make things sound as good as they can sound given where we’re at and the
13:29
equipment we’re using. We’re not striving for this ideal that, you know, if we hit the ideal, we might make
13:35
things sound a lot worse. Yeah. Or we might lose the objectivity
13:40
because then how does it transfer to a different environment, which is the whole goal here, the whole point. Yeah. Yeah. And there
13:46
is really good science that backs up if it sounds good, it should translate well
13:52
theoretically. Theoretically. Yeah. That’s always the in terms of a system, not in terms of a mix of course, but yeah, if your system
13:58
sounds good, it it ought to translate well. Um, and that’s
14:03
there’s a lot of research that really heavily supports that. So, you know, if you walk into a room and you listen to a
14:08
mix that you’re familiar with and say, “Holy crap, this sounds great.” Great mix on that system.
14:14
Yeah. So, how do you go about voicing these so that they are as neutral as
14:21
possible? All of those design philosophies that I just talked about in terms of um making
14:27
sure that the speakers are playing well in a room on access response in an anooic chamber should be totally
14:33
neutral. Um now there’s physical limitations around that and especially when you’re dealing with a budget speaker like our LP series, you’re not
14:40
going to quite achieve that standard. Um, when you’re dealing with a much more expensive speaker like our SM series, we
14:46
can aim at that standard. Okay. Um, and then making sure that the directivity of the speaker in a room is
14:54
clean, is linear, um, so that when we put these in a the general room that
15:00
they’re going to go, we can trust that at least translation above 700 htz where
15:05
the room doesn’t have as much of an effect is going to be valid. Mhm. Now you have this issue because
15:11
below 700 htz the room’s doing all sorts of funny stuff, right? Um and you can certainly, especially in
15:20
stereo, you can mix around it. You can learn where the gaps are in your
15:25
understanding based on what’s happening with the speakers in the room. Okay? You can learn to compensate for it.
15:31
Sure. Um, when you have typically when you’re working in stereo, even in an
15:37
asymmetrical room, left and right are going to sound more or less the same. Um, when you start talking about an
15:44
Atmos system, um, the same speaker at all these different spots in the room
15:50
can sound wildly different. Well, and not just that, but I would imagine that as soon as you stick
15:56
another speaker in the room, it’s going to change the sound of the existing speakers, right? I mean, because Yeah.
16:02
Yeah. And and I don’t mean that, you know, necessarily change the sonic character, but I mean, then you’re going to start having
16:07
different reflections, right? You’re going to have different interactions between all of the
16:13
speakers, different phasing issues and stuff like that. So, by the time you get up to, you know, a 7.1 system or
16:19
something, all bets are off. You could have chaos, right? Yeah. And especially from a mixing standpoint, you can’t have that.
16:26
Um because now we’re not just talking about a stereo image. talking about objects that can be placed anywhere
16:31
around the listener. And if a speaker isn’t behaving nicely with the other speakers in the system, you’re not going
16:36
to be able to hear that object placement correctly. And you know, something that you thought was a really cool artistic
16:42
choice is not going to translate to the end user, right? And then the other thing that I think uh it impacts as well, I would
16:50
imagine, is how does that mix then collapse down to stereo, right? you know, and
16:55
that’s something that I mean I know from talking to people like Matt and Bob Clear Mountain and others who do a lot
17:02
of Atmos mixing that is probably one of the biggest challenges. I mean it’s fun to play with the objects and everything
17:08
but then how does that how does that then turn into a decent stereo mix without losing it all
17:14
right and so uh what we are doing sort of around that um we have been using a
17:20
measurement technique called the moving microphone method um which is a fast measurement technique um fast reliable
17:28
accurate okay um so other measurement techniques um that have been used in the professional
17:33
sphere have included some array of microphones, right? Um, and the fewer microphones you have,
17:41
the easier it is to get different data. Um, so if I’m measuring the system with one microphone and I don’t like what I’m
17:47
seeing, I just move the mic and I will get as opposed to like the ambisonics kind of thing.
17:52
But even an ambisonics sort of thing is not capturing enough data and not necessarily averaging it correctly.
17:59
That’s what I wondered. And then there’s all the matrixing within that and how accurate is that? And so it’s another whole can of worms.
18:05
Yeah. So the moving microphone method um came to us via some people who were
18:11
tuning cinemas um because this is what they were using. Um because you have this giant space,
18:17
how do you make it as good as possible in as many seats as possible? And if you’re talking about microphone arrays
18:23
now, you’re really in trouble because you would need thousands of microphones to Right. because you’ve also got a
18:29
different a different perspective for every single seat in the theater. Yeah. So, the moving microphone method, what
18:34
it does is it captures a volume of space that is roughly equivalent to the area
18:39
occupied by the listener’s head. Um, and if you’re going to talk about a cinema space or a larger space, you can start
18:46
to capture more space than that. But for a music studio where there’s typically going to be one person mixing at the mix
18:52
position, we capture a volume of space that looks like the listener’s head. Okay. Um, we have very good data that
18:58
supports that that measurement method roughly correlates to several hundred
19:03
microphones at random places in the listening position. Really? Yeah,
19:09
that’s interesting. Um, and how that was captured is we took the same measurement several hundred
19:15
times, moving the microphone each time. Wow. Um, and then correlating that to the
19:20
moving microphone method. The moving microphone method gives you the same data. That’s some serious math. It’s some
19:26
serious math. Yeah. What’s powerful about this is I can take that measurement in about 30 seconds and we
19:33
always take it a second time to confirm that, you know, we we got a valid result the first time. Um, but we’re talking
19:40
about a minute per speaker to measure. And I have very reliable results.
19:46
Wow. I can take that measurement. I will get one result. You can take that measurement. And as long as you follow
19:52
what I’ve said in terms of moving the microphone away from your body so you’re not getting body reflections and really
19:57
taking the time to move it through that volume of space, you will get the same result.
20:03
Um whereas some other measurement methods, I can take a measurement, I get one result, you can take a measurement,
20:09
you will get a completely different result, which is what I would kind of assume. Sure. Yeah. Yeah.
20:14
The other thing that’s powerful about it is I can come in and take that measurement today. I will get one result. I can come in and take that
20:20
measurement three weeks from now and as long as nothing’s changed in terms of the speaker placement in the room, the furniture in the room, that sort of
20:27
thing. I will get the same measurement again. Now, let me ask the loaded question, of
20:33
course. What if other conditions have to do with it? For example, like humidity, stuff like that. It doesn’t seem to affect it in that big
20:39
a way really. Yeah. Now is that because it’s such a small space or because I know that you know for example um I’ve spoken to
20:47
people like Robert Scoville mixes in the sphere okay and he has been very adamant
20:53
about the fact that humidity is a huge huge factor because it especially
20:59
impacts your high-end and how the transmission of those waves is impacted
21:04
by you know not just the relative humidity itself but then you know you fill up a room with bodies and that’s
21:10
going to change the humidity and everything. Yeah, that’s obviously a much different sort of set of problems than what we’re
21:16
talking about here. But I’m wondering, I mean, does the relative humidity of a space not impact
21:22
that measurement? It doesn’t seem to in a statistically significant way. Interesting. Is it just because it’s a
21:27
smaller space or I don’t know if it’s because it’s a smaller space. I don’t know if it has to do with, you know, proximity to the
21:32
speaker. um uh the sort of listening levels we’re talking about um when we’re
21:37
talking about safe mixing. Um the comfort of the person sitting at
21:43
the mix position, you know, if the humidity is off the charts and they’re sweating, they don’t want to sit there for 8 hours.
21:49
Yeah. Exactly. Exactly. Um but it’s just that that’s interesting to me. I I would imagine it probably has
21:54
most mostly to do with just the size of the room because when you’re talking about I mean here you’re talking about
22:00
distances of just a few feet right at most. So that’s probably that’s
22:05
probably a really negligible difference in terms of transmission rate for those
22:10
high frequencies loss and everything. Yeah. Mhm. But no, the only reason I asked him is because you mentioned, you
22:15
know, doing it um getting that information from people who are doing theaters and I would think that at the
22:21
point that you get into, let’s say, a 500 room space, that would impact that, would it not?
22:26
Yeah. What they were finding consistently was that they could go in, take the measurement, and they could go in again some large amount of time
22:33
later, not the next day or a couple hours later. We’re talking weeks or months, and their measurements that they
22:38
had taken earlier still hold up. That’s interesting. I mean especially because you would then have so many
22:44
different so many more variables in that kind of space and again you know if is
22:50
is the auditorium full is it empty you know I mean that’s got to impact the yeah and I can’t speak to you know how
22:56
many bodies were in the space how they were for that all I can speak to is and you do not get this with a with a
23:02
microphone array you go in one day you take a measurement you go in some large amount of time later you take the same
23:08
measurement you get the same result. That’s really interesting. Why do you think that is?
23:13
Um, I think with microphone arrays, we really don’t give enough credence to um
23:19
how finicky placement is in terms of the result that we’re getting. Um, especially where, you know, if we’re
23:26
doing this in a in a purpose-built scientific environment where the microphones never move and
23:33
there is a very precise placement of whatever the source is, not such a big
23:38
deal. If we’re doing this in a movie theater where we are bringing in the mics and we are setting them up on stands, how do you know you’re exactly
23:45
where you were the other day? Right? Even if you’ve taped down a marker, it’s still, you know, it could be just skewed, you know, a quarter of
23:51
an inch or something like that. And that’s, you know, one of the things that I learned in recording something like an
23:57
orchestra is that, you know, literally like, you know, moving at a millimeter at a different angle is going to
24:02
radically change things much more than you expect it would. Yeah. you know, um, and that’s the science bears that
24:09
out. Uhhuh. Interesting. Okay. So, how do you I mean, I don’t know that we can accurately demo it in a you in a YouTube
24:17
sort of situation, but I’ I’d be interested to know the procedure of how you would measure something like this.
24:23
Yeah. So, we use Room Wizard. It’s free software. Um, there are other softwares that are capable of doing this. This is
24:29
there’s a bit of a learning curve for Room Wizard, but it’s free. It works very well. It’s super reliable. Um so
24:36
once everything is calibrated um so you have to calibrate your interface you have to have a calibrated microphone it’s not the typical microphone
24:43
calibration is usually on axis and then some series of off-axis calibrations right we are calibrating for random incidents
24:50
so we need to know that the microphone is accurate uh with what we’re doing we’re pointing that microphone in all
24:56
different directions while we’re taking the measurement um so we need that calibration on top of
25:01
things but you’ve got all that you’re off to the races the speakers are placed. Um, when you start taking the
25:08
measurement, uh, you’re basically just capturing an RTA with pink noise. Um, or you capture 100 measurements in
25:15
33 seconds. So, it’s three measurements per second. Um, and based on that, you
25:21
usually have the picture you need to start making decisions. Um, like I said
25:27
earlier, we always take that measurement and then immediately take the measurement again just to confirm. Sometimes there’s outside noise,
25:33
sometimes a mistake was made. Um, all sorts of things can happen. Um, but 90%
25:38
of the time we take the measurement, we take it immediately again, they’re right on top of each other, we move on to the next phase. Um, that next phase can look
25:46
at a lot of different ways. If we are in a in a studio build process, we can use
25:52
the data we’re capturing there to say, hey, we’ve got giant modes in all these different places,
25:58
right? So, fix it now. Fix it now. Let’s try moving the speakers elsewhere in the studio. Um, let’s try changing the listening
26:04
position. Let’s see if we need to invest in some bass traps. Um, let’s get that done. If the speakers
26:10
are say uh, you know, screwed into the walls and the listening position is exactly where it needs to be because we
26:16
cannot change that, then we start going about that with DSP um, and correcting those problems in digital EQ.
26:24
Room treatment, I guess, would be an option too. We hope. We hope. Yeah. Yeah. Um and at the end of the day, especially
26:30
when you’re in a smaller room. So, we’re in like a 10 by6 room right now. Yeah. Um you can only put in so much
26:38
treatment. Sure. You’re going to have you’re going to have modal issues no matter what you do.
26:44
So, finding a place where the speakers are behaving okay. Um and then correcting in DSP is not a problem at
26:51
all. And that seems to be endemic of what we’re seeing more and more of lately is smaller and smaller mix rooms,
26:59
you know, and it’s not just I mean, yes, people are putting them in their bedrooms and stuff, but even beyond that, you know, I’m seeing a lot of
27:05
studios now that are saying, “Hey, let’s put in an immersive room. We’ve got this little closet over here, you know, kind
27:11
of thing.” And it’s it’s always that way because, you know, like the main room which was built a long time ago, no,
27:17
we’re going to keep that as stereo, you know, because we don’t want to mess with what we’ve already got, right? So, let’s find a little niche and build an Atmos
27:24
room in there. And you get these Atmos rooms that are fairly tiny, you know, and I mean, that makes me wonder, too,
27:31
because again, how’s that sweet spot going to translate to another place? Yeah. And that’s where it becomes really
27:38
important to be able to the mix engineer needs to know that what
27:43
they are hearing in that space is accurate. And that’s not just a question of speaker choice and room design and
27:49
room size. Um even if you are not doing anything corrective in DSP
27:54
the mix engineer should be able to confidently say I am getting accurate frequency response from every speaker.
28:01
Uhhuh. Everything is in phase. Everything is time aligned. Sounds like a nice ideal.
28:06
Yeah. Um and all of that is at least theoretically possible without any sort of uh digital wizardry.
28:14
Um it’s nice to know though. It’s nice to have the measurements to back that up. Hell yeah. And once you’ve taken the measurements,
28:20
if you are going to do that in the digital domain, it’s really easy to do. How so?
28:25
Uh because you have everything you need. So you have a frequency response, how the speaker is playing in space.
28:33
to get it
28:39
and of course at that point you’re really getting into a lot of DSP. You’re getting into a lot of DSP. Yeah.
28:44
Um time alignment once you’ve learned the process is fairly trivial.
28:49
Yes. Yes. And that’s really I mean it’s done in almost every room these days.
28:54
And yeah phase shouldn’t be a problem if everything has been wired correctly. more wondering about face, you know,
29:00
because and I mean it’s it’s it’s almost a shame to say, but you know, I’ve heard
29:06
this so many times that, you know, you get this great system and then just,
29:11
you know, somebody has messed with the phase or somebody has not compensated for a phase correctly. And that’s
29:18
something that can go undetected much more easily than time alignment,
29:24
you know. Yeah, it’s a tricky thing. Um, if you have the measurements, you can just look
29:29
at the impulse response and see if the speakers are in phase with each other. Um, you’re done. In theory. Yeah.
29:35
In theory. If you don’t have the measurements, you are relying on your ears to pick up uh what can sometimes I
29:41
mean sometimes it’s it’s hey, that thing’s out of phase. Sure. Um, but then it’s then it’s painfully
29:47
obvious at that point. Um, but sometimes it’s not so obvious, especially when we’re talking about an immersive system, especially, you know, the speakers in
29:53
the back, the speakers in the top aren’t necessarily providing a lot of uh easily
29:58
identifiable content, right? Because there’s not that much content in there. It’s mostly ambience
30:04
and stuff like that, right? Um, so if that’s out of phase, you might just think that the sound designer thought everything ought to be over
30:10
here. Yeah. Um, or you might just think that that’s the way it should have been, you know,
30:16
mix-wise, you know, I mean, you’re just listening to something. How do you you know and and that’s one of the things that I think is so tricky about these
30:23
immersive environments is that the program material that we’re listening to and and that’s why I think it’s
30:29
interesting that you use so much uh science and math because so much of what
30:37
we’ve done in terms of designing rooms and building rooms and everything is I
30:42
don’t want to say it’s a black art because it’s never been a black art but so much of it comes down to your ears
30:47
and the idea of actually using math, dare I say it. Sure.
30:52
You know, is it’s almost a novelty in a sense. Yeah. There’s there’s two things that I
30:59
would point out. The one is that, and I said this earlier, everybody’s got
31:05
pretty good ears. They might not be able to tell you what’s going on, but they can tell you something’s not right.
31:10
Yeah. Um or they can at least tell you, “Hey, I enjoyed this more when you played it earlier.” I think that’s, you know, and
31:17
especially when you start talking about the lay person, you know, and and now, and of course, that goes back to a much
31:23
bigger argument, which is the argument that we as the mixers in the room, we’re
31:29
hearing something that no one else is going to have the same experience with. Sure. No matter what. And that’s always
31:35
been, you know, that’s been a point of frustration forever. you know, we we agonize over this great bass response,
31:41
over, you know, all this separation, everything, and then somebody puts it on even before we had earbuds, right?
31:46
Somebody would listen to it on, you know, a $49 stereo and sounds like right? You know, so understanding how that
31:53
translates, you know, that’s why NS10s and became so popular. Yeah. So, there’s there’s interesting
32:01
data around translation. Um, we know that a speaker that sounds good
32:08
usually translates well and vice versa. Yeah. Um, a speaker that sounds good is
32:13
neutrally representing um the program material. Mhm. Um,
32:18
we also know that if we average every speaker ever made, um, the crappiest car
32:24
speakers you have, um, $49 stereos, $10,000 hi-fi systems, those will
32:31
average to more or less neutral. Okay. Um, yeah. And so if if you want your program
32:39
material to sound as good as possible to as many different people as possible, your goal should be neutral.
32:46
Yes, that is the ideal. Yeah. Um, and the
32:51
I take umbrage with the speakers that intentionally use a wonky frequency response to sort of represent that you
32:58
know what it’s going to sound like when it sounds like right? Cuz no two crappy speakers sound
33:03
anything alike. Well, and not just that, but then you know if you start getting into that concept like you know I used to love to
33:10
walk outside in the hallway of my studio and listen to the mix, you No. Do you know what you’re hearing is sound power?
33:16
Yes. Exactly. Exactly. And I think that’s important. You know, I think it’s more important to be able to get out of
33:22
the field and listen to it for that sound power than necessarily sit in the middle of a pair of tones. Yeah. Because
33:29
all I’m doing then is just hearing the same thing with a different mix of frequencies. You know, am I really
33:35
getting an accurate anything? You know, that’s why I always hated NSt. Yeah. So, it’s funny that you mentioned
33:40
leaving the room because that’s actually something that we do in the design phase. How well is this speaker working
33:46
with the room? Yeah. You really want to know, go outside the room. Yes. If the mix holds up, the speaker’s doing
33:52
great. Yeah. Yeah. Exactly. Because typically if you go out there you’re going to
33:58
you’re going to lose certain things but if you’re still hearing to me I feel like if I’m still hearing the overall
34:06
timber the you know basically if I’m getting the and and again this is me using my ears.
34:12
math, but if I’m getting the vibe, you know,
34:18
honestly, that’s what we do for that vibe. If I’m out in the hallway,
34:23
then, you know, that’s a good sign to me. Yeah. Yeah. The other thing that you
34:28
mentioned with Blackmagic and with, you know, some of the sort of pre-math um
34:34
studio for um I think I can’t help it. I come from that era,
34:39
you know. Sure. Yeah. Yeah. I think part of the flawed thinking there is this idea that any one factor is more important than
34:46
others. Um, we get asked the question, Charles gets asked the question all the time, what is the single most important thing
34:52
when you’re designing a studio monitor? And the answer is how well does the entire system work as a system. It’s the
34:59
single most important thing, right? Which is not a single answer, which is not a single answer. But it’s not down to frequency response. It’s not
35:05
down to directivity. It’s not down to group delay. It’s not down to distortion. It’s down to how are all of
35:11
those factors working in concert with each other to accurately represent the program material. Yeah, makes sense.
35:17
Um, and the same thing is applies to studio design. It’s not about the size
35:23
of the room. It’s not about uh the shape of the room or the materials that are in there. It’s about all of these things
35:29
working together to make a good studio. Mhm. Um, and yeah, I think where I get
35:36
annoyed with snake oil and black magic is when it’s like, I’ve got it solved. This one problem I know how to solve
35:43
better than everyone else. It’s like, well, yeah, but you forgot about this, this, this, this, and this. Right. Sure, that tweeter sounds incredible,
35:49
but the speaker’s got no base response. Right. Or just, you know, yeah, I’ve got it solved. Now, let now let’s take that
35:55
into somebody else’s room. Is it still a solution? Right. Right. Because that’s the point that I think you know
36:01
everybody forgets is that you know you literally every single thing is going to change when you change one piece of that
36:07
environment let alone when you take it the whole thing into a different space. Yeah. And I would I would assert pretty
36:12
strongly that when you have a loudspeaker that is designed well as a system and that it doesn’t apply you
36:19
know more or less priority to any of these one factors that’s going to work in the widest possible variety of of
36:26
environments. Mhm. Yeah. Yeah, that makes sense. So, is there anything that you can demo in
36:32
terms of how you voice these things? Um, yeah. I mean, I can take a measurement of the speaker um without
36:38
any processing and show you. Yeah, I’m curious to see what that looks like, how the speaker is interacting with the
36:44
room and then we can apply the processing and take the measurement again and show how the EQ can sort of help things
36:49
because obviously this is not going to translate to a YouTube video, you know, in terms of somebody being able to listen to it sonically, but I think if
36:57
we can show what those measurements look like, that that sort of backs up your
37:02
theory, so to speak. Absolutely. Yeah, let’s do it. Yeah. Show me some math. I’m going to do math at you
37:09
aggressively. So what are we using here now? We are using roomq wizard
37:14
um and so we’re going to take a moving microphone measurement of this speaker um in its position.
37:22
Um and then we’re going to apply the correction. Um so what we do here we have this
37:29
earthworks microphone for which I do have the um the random incidence
37:34
calibration that I told you about earlier. Um, we’re going through this Volt interface. You can use any
37:40
interface of reasonable quality. Um, out to the speaker for purposes of time
37:46
alignment, which we may or may not do right now. Um, we also have a loop through coming out of the other channel
37:52
and back into the interface. Okay. So, that gives us a reference point so we can say exactly how much uh delay the
37:59
speaker itself is experiencing. Um, yeah. So, I’m going to turn the DSP
38:05
in that speaker off. We’re going to blast it with pink noise. We’re going to take a measurement. Um, and we’ll go to
38:11
next steps from there. Do it. All right. And here we go with an RTA. Um, because we’re just doing this for
38:18
demo purposes. I’m not going to ask you to move. Typically, um, I would have you sit somewhere against a sidewall and try
38:26
to be acoustically inert for me. I’m really good at being acoustically inert. Yeah. Um, so, okay, we should be ready to go.
38:32
So, I’m going to take an RTA. Um, I have this all set up. This is all available at caliaudio.com-santa
38:40
monica howto. Um, and this is all agnostic to any speaker you might be
38:45
measuring. So, obviously we’re we’re measuring these. This is on our site specifically to be used with these. Um,
38:50
but this process holds up for anything you might want to measure. Okay. Um, and we go through how to set all of
38:56
these various confusing settings so you get the valid result that you’re after.
39:01
Um, so we start recording and then we go to the generator and we start playing
39:06
pink noise.
39:11
I dig the hand motion, by the way. The the hand motion’s key. So, what we’re doing here is we’re just trying to
39:17
make sure that the microphone gets to face all different directions through that volume of space. Um, and I hold it
39:24
out like this so that reflections from my body aren’t necessarily uh influencing what happens. If I’m real
39:30
close here, that’ll happen. Um, especially if I uh stand to this
39:35
side of the speaker, I’ll start to get some reflections. So, Oh, sure. Okay. Just a little bit of intentionality
39:41
thinking about I don’t want to be again I want to be acoustically inert, right? Um, obviously there’s going to be some
39:47
bounce from me to the speaker, but it the data shows it’s fairly minimal if you position yourself right. Yeah,
39:53
makes sense. Okay, so we have our first measurement and I’m just going to take that immediately again to make sure that uh
39:59
the next measurement lines up. Make sure we didn’t capture some sort of extraneous noise or get an anomalous
40:05
reading here.
40:11
See that first measurement’s right on top of the second one. Yeah. Um, remarkable actually. So, from here we go
40:19
into room EQ wizard screen and I can use either of these two measurements. I’m
40:25
just going to get rid of the first one that I did. And you can see above 200
40:30
hertz, we’re we’re fairly consistent here. Um, if we use 72 as a baseline,
40:36
we’re up to 74 on the high end and down to 65 on the low end. Um, with the slope, that might not be so bad. So
40:42
that’s a reasonable window. We do have these huge modal issues below 200 Hz. So
40:47
we have a suckout right at 200. We’ve got a giant suckout between about 40 Hz and 150 Hz.
40:54
Um so what we’re going to do now is we’re going to use the autoeq function
40:59
on roomq wizard which is a handy handy function. Um, Cali audio speakers are
41:05
already built in here, which will help to control how many bands of EQ are assigned and how uh big the steps can be
41:14
based on what the speaker itself can handle. Um, importantly, we are going to build
41:19
in uh a slope here. Um, so we have a 3dB slope that seems to hold up fairly well.
41:27
Um, I might go up to 4 dB and see if that lines up a little better. So, I’m
41:32
actually going to use that point4 dB slope. And then we go to the filter tasks
41:38
and we have the computer try and match things to a target. Um, and you can see,
41:47
especially if we get rid of the measurement itself. Um, this should work really, really nicely. Um, this is
41:53
lining up really well. We can look at the EQ filters themselves. We do have a plus 15 which I don’t love at uh 95 and
42:00
a half htz. Although we also have at 97.3 hertz a minus 7. So that’s one of
42:06
those silly steps where we have this big thing but then we have something immediately next to it that kind of
42:11
takes it out when we’ve got a little notchy do. Um that can be helpful. Um the one area
42:16
of concern I have here we are going to hear this dip from about five to about 13. So, I’m just manually going to go in
42:24
and put in that um last filter. We’ll do that right at 10K.
42:32
So, that’s looking pretty good in terms of frequency response. So, what I’m going to do here is I’m going to save
42:38
this file and I’m going to save the
42:43
predicted result. So when we load the file into the speaker, the measurement
42:49
that we take afterwards should sit directly on the predicted uh result. Okay. Um that’s how we know everything’s
42:55
working. Um usually it works. Um okay. If we have uh problems that are in more
43:03
than one dimension, usually when we’re talking about room modes, the issue takes place in one dimension and it’s
43:08
not as big a deal in other dimensions. Okay. So when we apply the correction,
43:14
the sound power is what’s affected. Um even though we are amplifying that issue in whatever dimension it’s it’s big in
43:23
by adjusting the EQ, we are putting it back into the speaker and the total sound power in the room will reflect
43:28
that. Makes sense. In issue in situations where all three dimensions are acting negatively on the
43:35
same frequency, you’re out of luck. Move the speaker. But that’s something that this will help
43:41
to show. And now you can see what we’ve loaded in here. Mhm. Um, so that is now living in the
43:46
speaker. So let’s take the measurement again and make sure it’s what we expected it to be.
43:53
You can see we’ve got a little bit of weirdness at 40 Hz. We’ll redo this measurement to see if that goes away.
43:58
Um, but everything above that we are lining up pretty close. Yeah. Yeah. Wow. That’s interesting. Okay.
44:06
Um, so let’s take the measurement one more time. see what happens.
44:11
So, same thing here. Something’s going on between um 40 to 80 Hz that isn’t lining up with the prediction. So, from
44:18
here, if you know, if this were a mission critical um mixing room, we
44:23
would do some adjustments, right? We would probably measure the speaker again, make sure that measurement we got was valid. If the problem keeps showing
44:29
itself again and again and again, one of the things we would do right away is just raise the output level.
44:35
This could be a noise floor issue. Um, and so if we get the output level high enough, then we have a better signal to
44:41
noise and that might go away. Gotcha. Um, but there’s some troubleshooting to do here. What we can see is that it
44:48
worked really, really well over most of the frequency band. Yeah. Now, that’s fascinating. So,
44:53
basically, you’re able to load the load this compensation into every speaker.
44:58
Yep. And it just lives there permanently. Yeah. So, um, with the SM series, uh,
45:05
there’s two ways to do it. You can do online mode which we have in here where
45:10
every speaker is on the network and talking to the computer. Um we can load tunings in in real time.
45:17
We can make changes in real time. Um particularly powerful if you want to have two different curves as an example
45:24
because with a button press you can go between those curves with the two different tunings that you could load in. Now why would you want that?
45:30
Ah some people want X-curve some people want a Dolby curve. Oh sure of course. Okay. Right. That
45:37
makes sense. Or say you’ve got an artist coming in and they like they always want more bass
45:43
even though you know that’s a bad idea. You just have a button that’s just I’ve never had an artist like that.
45:50
Um yeah. So that’s one of the things that you can do with this. Um in offline mode you can load the tunings in over
45:56
USB. Uhhuh. Um and obviously that’s a little bit easier than running a separate wire to
46:02
every speaker. Um, one of the things you can do there as well is you can save, you can overwrite our tunings. Um, so we
46:09
have boundary compensation tunings based on where the speakers are in relation to walls and desks and other things like
46:15
that. Okay? You can use your own set of eight boundary compensations. Um, and so if
46:20
you were a studio owner, you could say studio A left, studio A right, studio B left, studio B right, and have a bunch
46:27
of floating speakers that the engineer can put up on there. Oh, that’s cool. Yeah. select from the back of the
46:32
speaker. They never have to use the computer to interface with it except for the first time getting it set up. Right. And they’ve got a calibrated speaker.
46:40
Now, that’s a nice touch. Yeah, I like that. Especially because yeah, a lot of studios have floating near
46:45
fields. Yep. Lots of stuff that you can do in the software. Um, I won’t bore you with with all of it, but if there’s anything
46:51
specific you’re curious about, this is really what I wanted to know about because this is something that,
46:56
you know, I’ve had these conversations with Charles before and uh, as I alluded to before, you know, talking to Charles
47:03
or talking to a lot of engineers, um, it’s uh, you kind of have to buckle in
47:08
and hang on for dear life. Sure. Yeah. Yes, because they will dazzle you with numbers and it’s fascinating. But for
47:15
those of us who are sort of in the real world, shall we say, you know, I want to understand it from a
47:21
conceptual p perspective, not a mathematical perspective. Because yes, I came up using my ears,
47:28
you know, and I want to know and and you’ve done a very good job of explaining how the numbers translate to
47:35
what I’m hearing. Yeah. Because that’s the point here, I think, isn’t it? It that’s Yeah. Yeah. I mean, the reason
47:40
we’re doing this and not, you know, selling drugs in Bakersfield is we we do
47:46
what we do to support art making. Mhm. Um, and at the end of the day, none of
47:51
you’re exactly to your point, none of the science matters. Does the song make you feel something?
47:57
Right. If we use the science and our understanding correctly, we can give creators a tool that makes it easier for
48:04
the song to make you feel something. Well, and see that’s what I’ve always loved about
48:10
some of the science when when when people approach it in the right way. That’s what I really love because you
48:15
know we are in a business that is such a weird mix of left and right brain
48:22
science and feel. Yeah. you know, and being able to say, “Okay, I can
48:30
I can replicate what you’re listening to and I can understand it from a mathematical perspective.”
48:36
there’s a certain confidence instilled in that where I can say to you, yeah, I can prove this with numbers, with
48:44
meters, whatever it is, you know, and then on the other hand, it’s like, yeah, you know, I’ve worked with a whole
48:49
lot of engineers who say, “Fuck the meters. I don’t want to see I don’t want to read the book, you know.” And I understand that. I understand both of
48:55
those perspectives. And I think at the end of the day, it’s
49:00
got to come down to can I create something that will work for that
49:07
intuitive approach. Yeah. So that’s what I I I I really like
49:12
this and and you’ve uh you’ve enlightened me quite a bit on this because that’s really what I’ve been
49:18
sort of questioning is, you know, okay, so how how does your geeky approach
49:24
actually translate to my real world music mixing experience, you know?
49:29
Yeah. And I think that’s I mean for us that is really the important question in a measurement technique. Is this is our
49:36
measurement translating? Yeah. because we could say with the microphone arrays sort of
49:43
um with some of the other measurement techniques that are popular. No. Um you know there’s a lot of times that
49:50
you will you will take a measurement and apply sort of an auto EQ and the whole process is automated.
49:55
Um and you get to the end and the speaker sounds worse. Yeah. Yeah. Um and that so obviously that’s not
50:02
good. Mhm. Um, we typically get to the end here, the speaker almost always sounds better
50:08
and then it’s an issue of can we try some different flavors? Um,
50:13
sure. Can we try thing? Yeah. Uh, yeah, it’s a it’s a personal taste.
50:18
It’s an experience thing, I think, more than anything. Um, somebody who is really familiar with their room will be
50:24
able to tell you better than I can with the numbers. Yeah. Uh, should we be using uh a more
50:32
high frequency bias slope or a more low frequency bias slope? Sure. Um, although they might not put it that way.
50:37
Sure. Um, I tuned Paul Third’s room in Scotland. Um, and I don’t remember we put in a 2
50:44
or.3 dB slope. So, what that is is from 20 to 20 there’s 2dB difference or 3dB
50:50
difference. And he said, it sounds the low end sounds a little aggressive and it sounds a bit dull. So, let’s change
50:57
it by 0.1 dB. Um, wow. And that’s what we did. Um, and he was
51:03
like, “Yep, that’s how it’s supposed to sound.” Now, did he actually ask for it like that? Did he know to ask that?
51:08
No. He He said the low end sounds, the high end sounds dull. Yeah. And I wouldn’t It sounded great to me. I
51:14
haven’t spent time in his room. Sure. Um, it lined up to the graph really, really nicely. Um, but he said, “Let’s
51:21
try something different.” We tried something different. It’s very subtle change and it made him happy. Interesting. I love it.
51:28
Yeah. Yeah. Fun stuff, huh? Very cool. Very cool. Well, Nate Baglios, thank you for being
51:34
my guest. Thank you. Thank you for the explanation. I really like this. Great. Cool stuff.
51:40
Cool. Thanks.
51:45
Hey, I’m Daniel Keller. Don’t forget to hit the subscribe button and join us each week for Insights and Sound.