top of page
  • Writer's pictureAdmin

Audio in the digital space



Welcome to Zeitgeist Radio. I'm your host, Morgan Roe, founder of the Zeitgeist Academy. Zeitgeist means spirit of the times, and it is the collection of cultural forces that all contribute to what it feels like to be alive and part of a dynamic culture. Every episode I speak with someone from a unique musical subculture.


We dig into their passion and explore how music is a powerful force that brings people together. Before we get into today's interview, make sure you head over to zeitgeistacademy. com and sign up for my newsletter. You'll get weekly mini lectures on cool musical facts, backgrounds on our amazing guests, and updates on what's going on with the Zeitgeist Academy.


I'm always up to something. Head over to zeitgeistacademy. com. That's Z E I T G E I S T academy dot com.​


**Morgan Roe:** My guest today is bridget holly an audio software engineer in the la area


Bridget. Welcome to Zeitgeist Radio.


**Brigit Hawley:** Thank you for having me.


**Morgan Roe:** Yeah. So for our folks out there, could you describe yourself a little bit? Who are you and what's your relationship with music?


**Brigit Hawley:** my name is Bridget Hawley. I am an audio software engineer. I grew up in a piano studio. My mom's taught piano for 50 plus years at this point.


And so from a very young age, I was surrounded by music and audio. And it really impacted my life. And even when I was in my undergrad in college, I started as a civil engineer. and I hated it. I got about a year and a half in and I was like, I can't do this. So unsurprisingly, I fell back into doing more music and relying on audio to like, bring me happiness.


And, um, I kind of looked at it and was like, is there anything I could do with a career here? I knew I didn't want to be a teacher. Cause I. Lived with one and I knew what that would be. And so I didn't want to follow it is a thing and I appreciate the teachers. Don't get me wrong, but it wasn't a path I wanted for myself.


So with that in mind, I also had always really loved video games. It was another one of the hobbies that I really enjoyed and still do to this day. And I had. You know, thoughts and dreams of basically working on soundscapes for video games and so I originally looked at doing like recording and stuff and that was fun, but it didn't like click for me.


I started doing audio programming, though, and that. And the rest is history.


**Morgan Roe:** look, but it's something that is literally, like you said, it's in our face every single day or in our ears every.


Single day, every day, at least for me, there's always music in my life every day. And I, you know, I don't have a clue how that happens, so I'm really excited for today. I heard that you actually created your own major. In college .


**Brigit Hawley:** Yeah. Technically. Yeah. How did you do that? What led you to do that? So when, so going back to the undergrad poly had the, the Cal Poly, San Luis had a pilot program at the time, which I think it's been made official since then.


But basically the idea of it was that you could take an. Engineering form and combine it with kind of anything else, but mostly geared towards arts. So it was called liberal arts and engineering studies. And the point of the program was that it was super flexible. And so you could really combine any engineering with anything you could think of.


I know somebody who kind of wanted to go into forensics. So they did biology and I think they were doing more of a computer science lead with that 1. And I know somebody else who wanted to go more into environmental testing. So they did environmental engineering and law was the other side. So like it was really a flexible program, which was really cool because there's not that many of those out there.


Yeah. It's like you have a goal in mind and you can kind of custom make what you want, the classes you want to take and set yourself up for a better career. And so that's basically what I did. I tried electrical engineering for a little while. It wasn't really my thing. That's going more back to the hardware.


Yeah. But programming and doing audio and a combination of music and game development courses is what led me to creating a major. Okay, hold on. So like, what


**Morgan Roe:** were some classes that you took


**Brigit Hawley:** for this? wEll you know, if you remember, well, so you had your basic like one on one classes and stuff, right.


Related to you know, all the one on ones, both on the music and computer science side, but when you got to the higher divisions, it was more like for the audio stuff I did recording. and mastering and mixing. I also did composition and a few other courses. On the computer science side, I did more of game development, so it was like computer graphics and game design and a few other high level classes like that.


Ironically, I also took a couple high level operating systems courses in computer science, which is more talking about how to Create and facilitate operating systems, which is actually where I ended up in a career. I'm better. I'm very good at it. I naturally, it comes easily for me somehow. I don't know why, but thinking in that manner I'm happy with where I ended up.


In the sense of like it's different than what I expected, but it's very fulfilling and I'm very lucky to be where I am. So yeah, nice.


**Morgan Roe:** Okay. Wow. That was, that's so cool. I cannot imagine knowing what like I wanted to do enough to create a major when I was


**Brigit Hawley:** 20 years old.


**Morgan Roe:** I had


**Brigit Hawley:** no clue. No, it's hard. Like my nephew is.


This is a tangent. He, he's struggling to find a major right now. And I just look at him and he's like 21 and I'm like, dude, it's okay.


**Morgan Roe:** I know when


I had to finally declare I was so upset because I'm like, can I just not keep taking classes? And like, why do I have to choose something? Why do


**Brigit Hawley:** I have to finish?


Oh, right. They charged me a lot of money to be here.


**Morgan Roe:** Yes. That means I should learn


what I want. No, no, I. That's just very, that's awesome. And then also, can you describe your master's degree? Because I think this is also super, super interesting.


**Brigit Hawley:** Sure. So my my first job out of college was working for an audio codec company.


And through a few acquisitions, they kind of changed and evolved into a newer company that had dreams of going into the Sure. So my first job out of college was working for an audio codec company.


And through a few acquisitions, they kind of changed and evolved into a newer company that had dreams of going into the machine learning space related to a variety of tech. And so when I learned about some of the things that they wanted to do, I started to imagine a future related to machine learning and audio.


And to me, that actually has a lot of potential still to this day. I know a lot of people think of things like Siri, or you know, most people are familiar with large language models, which are, you know, things that you can see for chat GPT. But machine learning has a lot of other applications. And so, for me, what I was thinking of is more personalization or things that would actually help accessibility for people related to audio.


And so yeah, my master's, I decided to go get a machine learning degree with the hopes of being able to look at the future of audio technologies.


**Morgan Roe:** That's awesome. So can you describe the difference between machine learning and AI? Because I thought they were the same at first, but they


**Brigit Hawley:** are not the same.


Right. They are slightly different. You can think of AI as being like, so artificial intelligence is the idea of creating an entire entity. Machine learning is more about being able to create an algorithm that can do one thing really, really well. So, like, you can honestly think of as being a brain and machine learning is being like a thought process.


So, like, , for example, if you have a machine learning algorithm, that's really good at identifying fruit. It's going to suck at doing anything else. You don't want to give it like language. You don't want it like it's not going to know what to do with it. It machine learning algorithms are generally very simplistic.


Yes, you can have things that learn and change behaviors and go certain directions. But again, that's because that's the thing you're trying to teach it to do. It's, it's like a verb, you can say a machine learning algorithm is the verb. The AI side is more of the, the entire capacity of trying to have like a, I don't want to say sentient entity, but you can kind of understand what I'm saying.


Something that can do everything and change its mind and react to different scenarios, if that makes sense. Yeah. So


**Morgan Roe:** machine learning is almost like muscle memory. Like you just, it does the thing that it's been trained to do a thousand times.


**Brigit Hawley:** Basically, yes, yes. And I mean, honestly, when you get down to it, machine learning is just.


Fancy statistics. It's really all it is. You collect a bunch of data and you train and test over that data. You know, there's a lot of different ways to do it, etc. But like fundamentally, it's, that is all it is. It's not smarter than that. It's not dumber than that. And that's why you can get biases really easily.


That's also why you get things like, you know, the problems we see with like chat GPT, where it's like really, really sure it knows the answer or something when it's like has, doesn't have all the facts or whatever it's cause it's the only data that it's been given and it's been told over and over.


Yeah, that's right. So,


**Morgan Roe:** so what are some of the ways that you saw machine learning or that you see machine learning impacting, like you mentioned accessibility, what are some examples of what that might look like?


**Brigit Hawley:** Well, so, like, there are technologies starting to come out that are related to being able to identify objects in a room, so to speak, like, using your phone camera or something.


There's studies that have been doing this for a while. It's not really new. But, like, let's say you took that. And combined it with, you know, if you had somebody who was visually impaired and you had something that could evaluate the surroundings for them in a room and maybe help them get around a new space that they haven't been to before.


Sure. That's an example. Also like there's other situations where you could think about having people, not just with impairments, but like, let's say there's a fire in your house or something. And like, you start to hear sounds and you're not home. Maybe your Siri or whatever Alexa at home picks up on the sound of the fire alarm it is able to send you an alert.


That's not necessarily accessibility, but it's another case where you could have something that's listening for audio and using it in a way to help create something that would. You know, improve your quality of life. And that's the kind of thing I think you could do with a lot of machine learning tech, just hasn't been entirely explored to the degree that I hope we will in the future.


**Morgan Roe:** , I know we're going to, you're going to have to like translate a lot of this down because you're really good at what you do. And I, I am around a little tech, but like, okay. So pretend I'm a five year old, how would you describe what you do?


**Brigit Hawley:** All right, so I could take this a couple ways.


I'm assuming that you want to know a little bit more about how audio works in the digital space. Okay. So At a high level, I, and somebody who's got a physics degree can talk more about the actual science behind it if you want, but at a high level, audio is wavelengths that travel through the air.


It's just waves and it's just stuff particles hitting, you know, creating stuff that goes through the air, right? So the way that recording works or anything that's transferred from the analog to the digital space is that you have a microphone or something that captures samples, which are basically the compression of the wavelengths.


And then well, so that's what comes in on the mic. And then what the Mike does from there, it takes samples. And so the samples, basically, you know, everybody seemed like the normal sine wave, right? That represents a lot of audio. So samples are taken periodically throughout that way. And so, what happens and when you start to get digital artifacts is if you don't have a high enough sample rate is you only capture pieces of the wave.


So like when you have this nice curve to start with, right, you start to get jagged steps going one way or the other because it's not You know high fidelity enough. And that's where the sampling rate stuff comes in, and that's why it makes a difference, is the more that you take, the more accurate you are to a sine wave, or whatever wave you're reproducing.


Sure. And that's how you can get and that's why also artificial sounds sound different than the real world sound, aside from the harmonics that are part of it too, which is another part of Again, physics would be better to explain that part. But if you just think of each sound as an individual wavelength, that's where the sampling rate comes in and why it's important to have the correct ratio, because the more samples you take, the bigger the file and that's the downside, right?


So that's how the data is converted. To something that the computer can record now when it's converted into something that it stores in memory, it is still converted to bits. So, you know, your zeros and ones. So that's something that also comes into it. So, like, when you see 32 bit or 16 bit recordings, it's related to how so, like, all those individual samples that I just talked about along that same way.


They're all mapped to a specific. Code, a specific sequence of bytes, bits and so like, if you don't have enough bits to represent the full wave, you're going to lose some more features of it, right? Whether or not that means you drop a sample or you compensate, like, there's a couple different ways you can do it, but you can kind of visualize, like, if you have a range that's too big.


And all of a sudden you can't retain those samples. It changes what it sounds like. Right. So all of that stuff is what goes into audio transitioning from analog to digital. So from there, how it gets to you as a consumer is slightly dependent on the content creators. And when I say content creator, that can literally mean anything from music artists to production companies for, like, movies.


Anything that creates and produces audio that you listen to, basically, you can think of, literally. Even Spotify, right? That's a form of a streaming platform. Right. And so when you publish something, generally people are somewhat familiar with different types of file formats. So like talking about your, your, your raw PCM is very rare to see transferred around because it, again, it's, it's uncompressed.


It's everything that's as close to the original as you could possibly get. What does PCM stand for? Pulse Compression Modulation, I think? I honestly, it's one of those things where you just say it enough that you kind of forget. Okay. You know what it is and you kind of forget what it stands for, but I'm pretty sure that's, it's something like that anyways.


**Morgan Roe:** But that's the raw, the raw data sort of, of the, of the sounds being put from the wave that I'm speaking into the computer that's recording


**Brigit Hawley:** this conversation. Exactly. Yep, yep, yep. Okay. And so that's the rawest, the closest thing you can get to the original analog signal. And so from there, generally speaking, the audio is compressed in some format.


There's a bunch of them. They exist, like, if you talk to different platforms, different content producers, I mean, Apple has their own, Microsoft has their own everything. Like I spoke about working for an audio codec company, that was one of the things that we did was write compression algorithms to take these bigger sequences of data.


tHere's a lot of


**Morgan Roe:** interesting, because I know that when, when we end this conversation, I'm recording on zoom. When we end this conversation and I ends this meeting, it's going to do this whole like compressing audio and it's going to, it's going to have this bar and I have to wait for it to compress the audio before it'll allow me to save it.


into


**Brigit Hawley:** a workable file. That's what it's doing, yes. It's compressing it into something that, so it goes through two steps. It's the compression, so that'll be one form, one layer, where it's basically taking all of the things it can reduce and putting it in a smaller, a smaller size. There's a


**Morgan Roe:** few, So the PCM file is enormous,


**Brigit Hawley:** right?


And then, Well, in relative speaking for audio, yes, yes. Right. Compared to video, no. But yes, right,


**Morgan Roe:** right. But it's, I'm just trying to picture like, so when it's compressing. It's taking, basically it's taking samples, right? Yep. It's pulling, it's using whatever algorithm has been given to it by Zoom to choose which pieces of the waves to keep how to,


**Brigit Hawley:** am I close?


You, you are very close. You can honestly, so there's a couple, there's different ways to do compression algorithms. Sure. But like a simplistic way you can think of a compression algorithm. It's basically trying to take. something where you have a lot of information and putting it into a smaller space. So, I mean, you can even think of something that's like a code sequence, like, you know, if you had something that was like let's say you had a sequence of letters that were like A, A, B, A.


You can take out the duplicate A at the beginning by saying there are two A's here and just translate that into another symbol that you represent, right? And then, so that's a, you'd reduce it down to three symbols at that point. You just have to say A squared BA. And so, like, that's a very simplistic way of thinking about it, but that's what compressions algorithms try to do.


Is they try to reduce the complexity of the bitstream into something that's smaller. And can be rehashed close enough back to the original there are so like when you hear things talk about like lossy compression, that's one of the cases where you will lose some of the fidelity so yeah talking about how you lose some of the accuracy or some of those higher frequencies that you know people don't necessarily hear or they don't really impact the sound as much.


That's another part that might get stripped out. So whenever you hear something talking about that kind of stuff, that's what they're mentioning. It's not so much that, by intention, the codecs and compression algorithms try to do that. It's just that we're trying to reduce the size. How do you do that without reducing the quality of the sound too much?


And there is a threshold. Like, phone calls and Bluetooth, people don't care nearly as much. compared to something that's like, you know, meant to be like a full on orchestra recording. You treat them differently. And so that's why there are different compression algorithms. Some do better for certain things than others.


And so because like a phone originally, you needed to have very, very small packets, right? To be able to have everything go fast. And that's also what happens when you have latency and stuff like Zoom. It's because You have all this data that's coming in if there's anywhere that there's a and going out to if there's anywhere in that pipeline that has, you know, not enough space or is too slow.


That's that's what happens. So that's why the compression algorithm really does matter. After the compression, it's going to be put into a container file, which people are familiar with like MP force. MP3s, those are container files. This is a way that you can contain, for lack of a better word, you can include like audio and video files or multiple audio tracks for the same video file.


That's kind of what that. Is the next layer and that's the package with which it is delivered to wherever it's going.


**Morgan Roe:** So is there a difference between those like an mp3 versus a wave? I'm never sure. It's like, which way, what do you want to convert it to? I'm like,


**Brigit Hawley:** I don't know. They are different. Yes.


Yes, they are. Wave in general is better for audio. But mp3s are also, they're both high enough quality for generally what people want to do. Like mp4s versus like mp2s there is a difference because like, one's meant for streaming, one's meant for basically like something you'd have on like a, a USB drive or something.


**Morgan Roe:** Yeah, so mp4s, isn't that video?


**Brigit Hawley:** That's what people think, but again, it's generally, it's a container which houses anything. So it can be video, it can be subtitles. Can be audio. Like, that's what I'm saying is there's the package, then there's the compression algorithm, then there's the original data. So, well, if you're going in the reverse order, so to speak.


But you can literally think of like MP4s or MPEG 2 streaming formats as the thing that's used to deliver the thing that you just compressed to your consumer. And anything that's done prior to that is interacting with the original data, if that makes sense. I think so.


**Morgan Roe:** So so let's go to Wave, and I don't know why this is different, but I've heard of lossless wave files I'm assuming that a lossless file is the opposite of what you just said, a lossy file.


**Brigit Hawley:** Yeah, so what that's talking about is the, again, it goes back to the compression algorithm. They're not going to strip out the higher frequencies is most likely what it means. I, I'm not, like, I don't know every codec out there by hand. Oh, of course, yeah. But what lossless generally means is it's talking about saying that.


Instead of reducing, taking out stuff that people don't normally care about, we're going to leave everything in.


**Morgan Roe:** Sure, so for music, that's probably important because of


**Brigit Hawley:** the harmonics. Generally, yes, but there are certain frequencies that studies have been done on that show that people really can't tell.


But there's, you know, there's always going to be people out there who feel differently. It's like the difference between like, you know, there's some people who are diehard analog fans or somebody who loves vital records. It's not that they're wrong. They just have that thing that they really appreciate.


You probably know this one. Audio people in general, gearheads especially, they get really, really passionate about how they feel about their audio and that's just what they like and that's okay. So going back to your question about music and generally, yeah, you want to use something that, and this goes back to which algorithm, which compression algorithm do you want to use?


Something that's lossless for something where you're like really care about the overtones is probably a better choice, but something for like a FaceTime call, not so much. Even Zoom, they generally don't have the same fidelity that you would probably want for like a full orchestra performance, so to speak.


Right. For sure. Yeah.


**Morgan Roe:** So cool. Okay, so file comes in, gets compressed. It's now in a package. Now what happens?


**Brigit Hawley:** So from there, it depends on what content it is. So like if it's on a streaming platform, it's going to be, the package is going to be some form of streaming. fiLe type and then so with streaming, it's a little different than a package file format that can be delivered on one, right?


Like the things that you can do in an all in one format is different than what a streaming situation can do because basically what happens in streaming is people want to start right away. They don't want to wait for the download, right? Right. And people get angry when the buffering bit is too long.


**Morgan Roe:** That's true. Yesterday, I was trying to pull up my Pandora, and it was taking forever, and I got really upset, and I just switched to YouTube.


**Brigit Hawley:** Exactly! So, what it's doing is that it starts to connect to a service hub, wherever it is, wherever So basically what happens, once you have that file, if it's going to be streaming, for example, it's posted somewhere digitally out there in the world.


And then you will, you as a consumer basically say, Hey, I want to access this file. And so a request goes out and, you know, a bunch of other stuff happens related to security and stuff that I'm not going to get into, but, and general internet protocol. But from there, you talk to the service hub and say, I want this piece of media.


And the service, I was like, cool, you're good to have it. And then it starts sending. Packages of, so basically what happens with the streaming format again is that package is is broken up into little individual letters, so to speak. And that's not really, that's not a technical term. I'm trying to more use it as a description rather than sending one big box.


We're going to send individual like. Tinier boxes along the way is the idea. And so each of those packets can only depend on the previous packets that have already arrived. Right? So you can't have it be like, so some compression algorithms and stuff use, especially in video, use things like I frames or whatever that are important for, future decoding. It's basically like this is a very audio doesn't really work this way as much but video. It's very important you have like your starting frame and then everything else is dependent on it. And so because audio and video are generally shipped together, because that's very common right and that's kind of like where we started with a digital again this is digital platforms I'm not talking like radio so much, because that's honestly not my forte I don't think I can really talk about that but When you're talking about streaming services, video and audio have been going together for a long time, like YouTube, basically, how old is that?


So yes, anyway, so with all those things being sent separately, none of you have to be very specific about the prior dependencies, or if you lost a package, like, for example, if you only got like you sent the service hub is going to send 100 packets. And you only got 98. Or something and you lost like two of them, your individual packets and compression inside of those packets has to be able to function well enough.


And so that's the kind of stuff that happens when you get like audio dropouts or frozen frames. It's because you've lost data. And the, the thing that's using it so like YouTube or Spotify is doing its best. To just kind of recover and let you continue experiencing whatever service that you are does that make sense?


**Morgan Roe:** Yes, it does. This is so fascinating. Where does hardware come into that? We thought you mentioned gear heads and. I know a lot of them. I am not one, but I know a lot of them. And like you said, respect that. And they will go on and on about sensitivities specifically of microphones or the quality of the data coming in.


How much do you interact with that? Or how much do you know about how, how much that actually. Effects the sounds that like, because it sounds like limitations can happen at any point in this process. Like, yes. Oh, yeah, I can have a better microphone, but it's irrelevant. If my or if the, what did you call it?


The container is not capable of receiving that high. Quality.


**Brigit Hawley:** So, so hardware definitely comes into this. Yes. So talking about the original data creation, which, which microphone you pick absolutely can matter. So remember talking about sample rates. Yes. And like, so it's not just about. So, okay. A microphone in a very simplistic sense is literally basically this thing that has a.


Magnet on it that moves and fluctuates with the wavelength and then those individual pulses are what's recorded and what's creates that again that the wave sequence that we talked about earlier. So different microphones. are better at different things, right? So like if we talk about being able to pick up different frequencies or, you know, have a bigger range of what they can take and even what, you know, what their max sampling rate is.


Like, that's all part of something that comes into this. Some of the, like, cheaper stuff and, like, it also even comes down to, like, the, believe it or not, the hardware going into your system itself, the connector cables, so, like, USB mics Firewire is an older format that not that many people are using now, but it was the big thing at the time for audio people.


So yeah, that stuff into like when you're talking about doing actual recording, you've got your nine pin connectors. You've got like all of this stuff comes into play in the sense of like how much of that that original wave can you capture and not just that wave, but whatever other stuff is happening.


So those overtones those especially talking about. or music like having the ability to capture all of it totally depends on and this is just the start This is just the beginning of the microphone. Yeah and that's why I mean this is also why digital sound like, you know stuff that's just made by a computer does sound different is because it's You're never gonna have the full sine wave experience or Outside mics.


I keep saying that. The, like, you're never going to have the exact wave like experience you do in real life, period. And that's why, you know, specific notes, like if you hit an A in certain programs, sounds like way goofier than what you can hear out in the real world, so to speak. sO that is literally just the beginning of this process.


And you hinted at is like The packets and stuff or whatever that's happening and there are so many steps along this whole way that things can change or be done, be done differently operate differently and also fail differently. And that's also goes back to how many audio codecs are there out there?


There's a lot. Let's just put it that way. How many different packet container types are there? There's still quite a lot. And then coming back on the other side, whatever's playing it back. There's also how many different hardware pieces are there? How many speaker types are there out there? Like there's so many.


**Morgan Roe:** Yes. And so cause that's a whole nother thing. Again, this is a podcast. I have been learning about, you know, audio and how, how fancy, I mean, you can gear head into it. You can spend a lot of money, but ultimately most people are probably listening on their earbuds or in a car or, you know, in a, from a device that Likely not super high fidelity.


I don't know this. Is that the right way? Exactly. Phone speakers. Yeah, yeah. Versus if I'm going to sit and listen to an orchestra record, like, like something we're doing Mozart this, this year, if I want to sit and listen to a really good recording. I'm going to play it on my really good speakers with my surround sound and all of that, like, that's a whole different, uh, expectation and


experience.


**Brigit Hawley:** Yes, exactly. And that is a hundred percent true. Part of that is down to the listener. This goes back to the gearheads. Yes. Sometimes gearheads really, really are passionate. Again, this is not a bad thing. It's just, that's... how they are. They're super passionate about, I must listen to all of my music on high fidelity speakers.


It has to be this way, which is totally fine. And in a certain extent, they're correct. Like there, there is definitely, and you, I mean, pretty much most people can say this. There is a difference between your phone speaker and sitting in front of like, a really nice pair of you know, Fairfield monitors that you have on your desk or whatever, or even if you go to like a movie theater, right?


Right. Those speakers are different than what you experience on a day to day life. So yes, in short, it comes down to what the consumer is expecting and what their trade off for the limitations versus the quality that they are willing to accept. anD that's also, I mean, even phone companies can get super into debating.


Oh, do we have the better speakers? Do we have the better camera? Like this is all part of it. It's kind of funny if you think about it from just a consumer perspective, because again, this goes back to why I'm passionate about what I do is that. Audio like digital audio is literally all around us every single day.


It's used everywhere. People just don't really notice it as much because the truth about audio is people don't notice it if it's working correctly, if it stops working, that's when they care, MARKER


Phew! How are you doing through all this technical stuff? I really hope you're able to follow along, because I found this fascinating. Oh, and by the way, did you know that Bridget is a master of Krav Maga and passed an absolutely brutal test just before our interview? She's a fascinating woman, and if you're not getting my newsletter, you're gonna miss out on some really fun side content.


Head over to zeitgeistacademy. com to sign up now so you don't miss it. Now back to the interview.


**Morgan Roe:** So what are audio. Drivers, that's a whole nother thing.


**Brigit Hawley:** Yes, so, this goes back to your question about the hardware.


So, remember how we were talking about there's a bunch of different microphones, there's a bunch of different speakers? Well, the truth is, each of those So, okay, the layer between Hardware and software is called a driver. That's true for anything. When you're talking about audio drivers, it's just related to stuff that's related to sound, so your microphone or your speakers.


And so, because of the fact that there are so many different companies and stuff out there that make different microphones or speakers, you have, and, but, your computer His only one. It only understands what it knows. And so you have to have a way for it to talk to these pieces of hardware. And that's where the audio drivers come in, is that you can swap them out, basically the audio drivers take what the hardware understands and converts it into something that the computer operating system can understand.


That's its function. So


**Morgan Roe:** this is, it's like a translator at the level of, so the, okay, just going back to, what did you call it? P. M. P. C. M. P. C. M. P. C. M. So the microphone's picking up the wave. It's putting it into this P. C. M. file is, is that what the driver then takes and translates?


**Brigit Hawley:** So, at the stage where the audio drivers come into play.


The data is generally already in bytes. So, okay, it depends on which way you're going. If you're going from the, the, the file format out to a speaker, for example, it's already in bytes. Okay. So basically you just take that data and for the audio driver's perspective, you basically write it to its buffers and then the audio driver itself is the thing that does the translation out to.


The hardware. Okay,


**Morgan Roe:** so you just said this word buffers and we've all had the experience where it says buffering. What is that?


**Brigit Hawley:** tHose are different things. Buffering, well, they are and they aren't. So, when you see something like in a streaming service that says buffering it's talking about hold on, we're going to go get more data.


Like, that's pretty much what it's saying. aNd it's kind of the same as what I say when I say buffers, is that basically there's a set amount of memory On your system that the hardware has access to, and the audio drivers that it can then take the data and so we were just talking about the file formats.


So, basically, what happens is the operating system takes the file takes it out of the package takes it out of the, compressed algorithm, so now it is that PCM, it's those individual bytes, bits that we were just talking about, and that is what's then transferred to the hardware. So, that is done via having this memory space that's basically like, here, put your memory, your audio data here, and I will now play it out to the speakers, essentially.


aNd that is the kind of work that I currently do for a living. is I where I'm the level right above the audio drivers where I'm taking everything from the rest of the system and pushing it out to the speaker or vice versa coming from the microphone to the rest of the system. So, you work with the drivers, the layer above it layer.


So, yeah, because you can think of the drivers. Your word of translator is perfect. Each driver is unique translator. I'm the layer above that that says, okay, I'm taking everything that's been converted to a language. My computer will understand and putting it out to the rest of the system. So, wherever it goes from there, maybe it's for you know, Spotify.


Maybe it's for FaceTime or whatever or zoom, whatever you're using at that time. That's the layer that in which I work in the operating system. So does that answer your Is all different? Every platform different? Yes. I mean, each operating system is different in fundamentally anyways. You know, that's why there's not, you don't see a ton of smaller brand names or smaller names for operating systems.


People do make them. Yeah. They're just not super common. And this goes back to usability and, you know, support across. It's a lot easier to have one operating system that you support or whatever. And this also goes down to the debate between, like, Mac and Windows or whatever is that one of the benefits of, like, Apple and Mac is that you have dedicated hardware.


And you have a dedicated software working together to create a certain level of experience that basically we're trying to guarantee Windows and stuff. The cool thing about that is that it's a little bit more flexible. You can like trade out pieces of hardware a little bit easier, be a little bit more customizable, but you lose a certain standard of it's going back to how much variety there is out there.


It's a little, Less consistent in what you might get at the end. You might get a superb product, but you also might get something that's not as great. Like, it's just, it's not saying it's a bad thing, like I said, it's more customizable in a certain way to whatever you like. So yeah, this all ties together, believe it or not.


I know it sounds weird, but it's true. It all comes together.


**Morgan Roe:** Absolutely does. So then let's go back to, so let's talk about audio snobs. Are you an


**Brigit Hawley:** audio snob? First of all? Yes and no. I mean, the thing is like, I know enough to know what it is and I care about it to a certain extent. There are certain things that are absolute pet peeves for me, but most of the time, like audio artifacts in shows.


Can really drive me up a wall, like if there's a show, like if there's a bad mix or something where like you have like timings off, like, have you ever watched something where like, sometimes that's actually because of the streaming service messing up, like that does happen, but there's also just sometimes where it's a bad mix, like maybe you have something where the voice is too low and the background is like completely blowing it, like you can't even understand what they're saying or whatever, like it's that kind of stuff, I don't know.


Stuff like that is kind of a pet peeve for me. I had this one time where I was watching a show with my roommate an old roommate of mine who's still one, still is my best friend to this day. She really loves the show and the audio cues for the scene that was coming up were like five or six seconds before.


And so it was like total peaceful, total peaceful. And then like horrible, everything's going bad music, but the screen hadn't changed. Like there wasn't anything yet. And then like five seconds later, it became like you know, something bad was happening and like, it made sense from the storytelling, but it was just like, that is a huge gap.


You have literally just made me, somebody who listens to audio, freak out five seconds ahead of the drop. And it was somebody's choice to do it that way. That wasn't, that was an actual mixer's choice. It wasn't the way it was. Somebody decided to make it that way. Right. So stuff like that does bother me.


Yeah. Yeah. But other than that, I mean, to me, the truth is a lot of the time is gear and stuff can get super expensive. Yeah. And there is a certain Yes, it can. There is a threshold. To me, there is a threshold of, like, how much is it worth it versus what am I getting out? Because, like Like, we were just talking about, a lot of people are fine with just listening to stuff on their phones.


Like, that's just how they are. And I mean, the truth is, most of the time, it's good enough at what it does that it'll get you going. It's just care, it's just how much you care about the other details past that. And so, there is some truth to it. I don't know that I'm that far on that spectrum, to put it that way.


Yeah, so


**Morgan Roe:** when, when you what types of things would in this process would be the most impactful for calling something like a high quality sound versus not


**Brigit Hawley:** in a certain way that's subjective, I hate to say, but like, it, you know, depends on the so like, One of the things the simple stuff you can look for is kind of what I started talking about at the beginning your sample rate Right.


That's what


**Morgan Roe:** everybody talks about


**Brigit Hawley:** sample rate. Yeah, you're big thing. It's one of the big things your bit length So going back to that how many? Individual samples, can you know? So that bit length, so, like, if you see 32 bit, it's pretty common now. Some people do 64 or 16 still. That's another thing to be aware of.


And then it passed that it's going to be probably specific on your individual needs. So like different compression algorithms cut out different things. So like they'll cut out higher or lower tones. And that's why you have different compression algorithms for depending on what you're trying to do.


The compression algorithm being used on Zoom right now, for example, is going to be different than something that you would have in like a DAW. Well, unless you use the DAW to have a plugin that allows you to Do that, but you can understand that the, the point of like a recording system where you're trying to do like music tracks is different.


And so the compression that's used is going to be different. Most likely. I mean, you could use the same, it's just, you're going to have a better quality experience if you find something that's more curtailed to your specific needs. Right. Yes.


**Morgan Roe:** I could literally talk to you for another hour after this, but Can we talk really quickly about.


Surround sound. I mean, we can even be just like super, super basic and just talk about like this pair of headphones that I'm wearing sometimes, like, how is it something with, I just have these two earbuds. How can I hear directional sound? Is that something you even work? It's something that just blows my mind.


And I'm like, how, how can I hear a sound coming from upper left hand corner of the ceiling? Like, like it's so directional. Absolutely. I just have these two, you know, it's


**Brigit Hawley:** amazing. It's amazing. Honestly, that's, that's one of the, what you're describing is actually one of the driving factors of why I wanted to get a machine learning degree, but I will leave that as a tidbit for the end.


So three dimensional sound, the way it works in real life is that everybody's ears are unique. Everybody's, they're basically, you basically have fingerprints in your ears, like, they're, your shape, the way it's built, everybody's ears are different, they really are, but there are things that affect what goes into your ears, so like the things that are in your room, the space, and the fact that, like, even if you have nothing around you, a sound, if it's like slightly over here, right sorry, I'm putting it right out in front of my right ear, is going to hit my right ear, geometry wise, right, It's probably going to hit my right ear before it hits my left ear, right?


So, that's actually how three dimensional sound works and why you can pick up a direction is because your brain receives this wavelength at different times. And so it's like, oh, okay, based on what's around me, and this is also why sometimes your ears can get tricked, is because they're like, what, where, where is it actually coming from is because all your brain does is it takes the timing of the difference between it.


Hitting one ear in the other and correlating it with what the your past experiences have been, so to speak. Wow. And this is, you know, go, I can't get into the, like, the biology of what your brain does, but, like, there's a lot more stuff there, and there's a lot more stuff in the physics world related to, like, textures and stuff like that, but, like, at a very high level, that's what's happening when you hear a sound and are like, oh, it's over there.


Okay, so then when you're talking about digital world, so there are a couple things you can do. The thing that's been around for a long time is something called like panning and delay. So panning is, so going back to the wavelength, you can pan to a specific. Speaker, right? Your right or left to say, I want the sound form or waveform to come out of this speaker or not that speaker or, you know, partially out of this speaker.


Like, you can do that. So that's basically what panning is and then the delay goes back to how quickly does it hit a year? Right? And so between those 2 things, that's the older way of trying to create 3 dimensional sound.


**Morgan Roe:** So can I take this and make sure I'm understand? So when I again, when I edit this audio, not so much with zoom, but if I have, an in person interview will have 2 microphones they're plugged in. I have a right and a left, right? So at that stereo, there's, I can either have it coming out of right or left, or I can convert the file to mono, and I'm assuming that just means equal, right? Half right, half left for all sounds.


**Brigit Hawley:** Well, it depends on which stage you're doing it in.


So some microphones, yes, have the ability to do right, left on them. And that's just how it's written in the file. These would be two separate microphones. Okay, so two separate microphones that are in the same room. So there's going to be potentially some background sound that comes in crosses. Yes, it's very annoying and I have to edit that out too.


That's actually why for recording, generally you put your musicians in separate boxes, actually, so that you don't have that. Pick up in the background. But related to if you wanted to. So basically what would happen is the, if you're creating like let's say we weren't in the same room, like we're here, and somebody got this file and wanted to make it sound like we were in the same room.


Then yes, basically we'd have our two mono tracks or whatever, most likely. And so, okay, Zoom puts it all into one file, I'm assuming, right, when you export it. Yep. In reality, those are probably two different streams, two different pieces of audio. Even if it's all compressed into one thing, like that's really what's happening under the hood.


So if somebody were going to take those individual waveforms, they could absolutely put one to the right and one to the left in the mixing process. That's where mixing comes in. So you've got your recording, mixing, mastering, production. Like that's another, that's a whole nother, that's a whole nother conversation.


Yeah. So like you can do that kind of stuff. And you could add that delay and that panning to make it sound like it's that by the, you know, multi dimensional sound. And so that's how that would work. If that was something you wanted to do, talking about removing the artifacts. If you're in the same room, it's the same thing, but in reverse is you have everything in the same.


All compressed into the same space or whatever and now you've got to try and have a way of using tools to extract it again that's technically the mixing process, even if it doesn't feel like it. So, yeah, that's how three dimensional audio has worked for a long time. In more recent years, and so, so, I should say this isn't necessarily super new yet concept of head related transfer functions have been around for, like, 50 plus years.


And so. The idea of limiting sound to specific speakers and delay and panning like that actually still has some downsides. Right? Like, the thing is that when you create a stream like that, it's fixed. It can't move around. And like, if you have. You know, people have different setups in their home, especially with radio, right?


Oh, yeah. Your speakers are at different positions. Your phone speakers different than like your TV speakers. Like, that's the truth of it. And so if your mix is set to. Operate in a certain way with panning and delay. It's baked in. It's permanent. Sure. So you can't change and better fit an environment to produce a better three dimensional sound.


And so it'll probably be something that you're like, Oh, yeah, I was supposed to kind of come from over there. But it's like. Not really right anymore, right? Yeah. So, the idea of HRTFs has been to more create something that's more flexible and unique to your ears. So, and there's still some controversy with that too, but HRTFs try to basically take sounds and model them based as objects in space.


In the the data and then allow those individual objects to be, uh, basically think of it like another layer. Yeah. Out of the compression, there's a way to say between the hardware and the system. What would be the position accurately for this if I wanted this to render at a specific location in space or as close as you can get.


That's fascinating. Yes. So basically, so


**Morgan Roe:** that would be a driver thing, right? A translator thing that the driver would have to


**Brigit Hawley:** translate. It's kind of everything You have to tell


**Morgan Roe:** the operating system. This is. The system, my


**Brigit Hawley:** configuration, this is my configuration,


**Morgan Roe:** 12 speakers, or I have three speakers or I have,


**Brigit Hawley:** so this is all together.


So the driver is. The part that goes directly to the hardware, but there's probably another layer above that even before me in the operating system. That's like, okay, we understand that there are 12 speakers in the room and the room is, you know, 10 feet tall, 10 by 10, whatever, like, you know, like that kind of thing.


And, from there, there's going to be different, there are more steps, there's more post processing, is what you can think of that happen here, to take that audio and make it so that those eventual bytes that are written from the operating system to the hardware itself are configured to create the sound in the space.


So what HRTFs do is they try to Simulate the same stuff that I was just talking about of like, what your brain does with sound where it's trying to figure out the location based on your head shape, your, your shape. And that's why the downside of is right now. Most algorithms use. A general form on and like I said earlier somewhere.


Yeah. Yeah. And like I said it earlier and this is why some, like, have you ever tried a, like an audio mix that like has been designed for three dimensional sound. I have no idea. I, I would recommend it. It's very interesting. It's how would you even do that? So, like, it goes back to you could have some that are done doing panning.


Absolutely. But something that's done with an HRTF is different and it's more accurate tends to be. It's also again, more flexible to be able to fit better for an environment, depending on your. Your codec that you're using your compression system. So these things all play together.


**Morgan Roe:** Yeah. I mean, I've listened to I've had very immersive moments with music where, where you, the, like, you'll close your eyes and you'll feel like you're spinning or like something is spinning around you, you know, like that's, I don't know if that's, that's probably not the tech that you're talking about, but that's an experience that I've had where it's,


**Brigit Hawley:** yeah, it's not quite the tech.


Let's go. But at the same time, what you're describing is a way of music kind of doing something similar. liKe if the let's say the composer wanted that to be the, the experience for the listener. I


**Morgan Roe:** can tell some, when I can tell it's intentional, which is most of the time, it's Not just like some like my speaker going out or something in one ear.


It's like, it's like a whisper will happen like slightly behind my, my right


**Brigit Hawley:** ear or something like that's probably done using panning. It's probably less likely to be trying to imitate your world. It's very HRTS are very similar to what your eyes or what 3d movies do where they're trying to like, You know, trick your brain into seeing something coming out.


It's very similar with an HRTF. And so, going back to it being a generalized form is the truth is sometimes they don't sound quite right. bEcause you're like, Your head is different from somebody else's. Like I said, your ears are in different locations. They have different structures. Yeah. And so going back to the machine learning tidbit I talked about earlier, probably the best way to actually have an immersive experience for people is to have something curtailed specifically to a unique.


So if you had a system that was well trained and smart and good, so a good machine learning algorithm that can take some user inputs or learn over time from some user inputs to create a more immersive experience where you have three dimensional audio. Just imagine the stuff that you could have for them to be able to play with.


You know, could affect video games, could affect media in general, movies. Yeah. And even just stuff like Spotify, if they, again, this does come back to the service or whatever, providing the ability to interact with this stuff, even if the hardware or operating system is capable of doing these things, if the content producers don't.


Allow for it, then it doesn't really matter. It's just going to go back to the default way of things, which is fine. Again, there's nothing wrong with it. Everybody functions quite fine with it today. It's just, to me, it's one of those things that I get excited about. Cause it's just thinking about the different ways that things can still be improved for media and tech and audio specifically.


I was a little rambly. I hope that still answers your question.


**Morgan Roe:** So interesting. Again, I could. Easily spend another hour on this, but we do have to wrap it up. I do need to ask you my final question. I ask every guest this do you know what Zeitgeist means?


**Brigit Hawley:** I think I've heard it before, but I also don't want to be misquoting the wrong thing.


So feel free to tell me.


**Morgan Roe:** A lot of people don't, it's a weird word, not a great business idea to name something after something most people can't pronounce, but. I'll explain. So zeitgeist is a German word. It means spirit of the times. And it's kind of that feeling of like what it's like, like right now with AI coming out, like there's this very specific feel where like excited and scared and like all of these things that will not happen.


Like this is a snapshot in time, right? There's a feeling for what it's like to be alive right now. And part of why, like a big part of why I'm doing this, this project at all is you can take that feeling and apply it to any point in history, right? What was the zeitgeist? Cause history is just this constantly evolving zeitgeist.


And if you can plug into, you know, even just thinking about audio sounds, you know, when, when records first came out, that would have been a really transformative time to have music in your home that you could play whenever you want it. Like that would have been a very specific feeling in addition to, you know, nothing exists in a vacuum as well.


So the politics of the day inform the zeitgeist, the gossip and the, the scandals and all of that informs the zeitgeist. So it's, it's kind of a, a word that refers to the feeling of what it's like to be alive at any particular moment.


Taking that to the music side, I've coined what I call a zeitgeist moment where you're listening to music and you just plug in and it could be music from today, or it could be music from 200 years ago.


But , it just. Comes alive for you and you feel like you're part of something bigger than yourself.


We've


all had that moment, so what was either a recent Zeitgeist moment for you or a particularly memorable one?


**Brigit Hawley:** I think I have two. One of them's not really music though, but it is sound. Yeah. Yeah. So the first, honestly, like, I think one of the first times I can't even remember which movie it was. I want to say it was one of the Star Wars movies, but it was definitely John Williams soundtracks. Yeah. And just having that like some of the like grandiose things that he puts into his music and the way that it makes like the ebbs and the flows like I have to say that was definitely something that like I can feel.


I can still feel to this day of being like connecting with the emotions tied to it. One of the other experiences that, again, going back to being fascinated with audio schemes and immersive experiences This is gonna sound weird. It's very specific. There was a horror game that came out in the 2000s. I think it was like 2008 ish 2009 maybe, maybe even 2010s called amnesia the dark descent.


And it's a horror. Adventure game where you're basically like you wake up in this like dungeon and you've completely lost your memory amnesia. There you go. And you don't know why you're there, but like, there's clearly something that's like, not, it's like Lovecraftian ask, like, it's a horror mystery kind of thing.


And like basically, as you're like exploring, you can hear stuff and like hear where things are coming from. And like you know. There's monster, monster, really monsters, whatever, things that are, like, trying to kill you, trying to find you, and so, like, the difference of it just being, like, a, like, if you, for example, if you mute that game, and you just play it, it's nowhere near as scary or as immersive as the audio that's tied to it, because you're, like, going through and you're, like, trying to, like, find You Stuff or like find clues and you're just like, Oh my gosh, this thing is going to catch me.


Oh my gosh, I gotta hide like that kind of thing. And you don't have that experience without the audio. And so like there've been other cases like that. I just feel like that one was a particular moment for me. So I always think of that when it comes down to something related to immersive audio.


**Morgan Roe:** Awesome. Love it. Well, Bridget, thank you so much for being on my podcast. I learned so much and I hope my listeners did as well. I really appreciate you and your time. Thank you.


**Brigit Hawley:** Thank you. I hope I was able to answer all of your questions. Well, we didn't even get


**Morgan Roe:** all of my questions. I have like an entire page left of questions, but you answered a very, very well.


The ones that we were able to get to today.


**Brigit Hawley:** I'm glad to hear it.


Thanks for tuning in to this episode of Zeitgeist Radio. To up-level your musical journey and become a music student for life. Join the Zeitgeist Academy by signing up for my biweekly newsletter. You'll get exclusive content, blog posts, and behind the scenes insights. I love putting it together and you'll love reading it.


Head over to zeitgeist academy.com/radio. That's Z E I T G E I S t academy.com/radio. Music for this episode was created by Ian Boswell. Please hit that subscribe button and tell all your friends you found a cool new podcast. See you next time.



2 views0 comments

Recent Posts

See All

Comments


bottom of page