Embedded - 507: Turn Our Data Into Predators

Episode Date: August 7, 2025

Chris and Elecia chat about books, courses, alternate podcasts, electronics, statistics, kidnapping Roo, and journaling failures.  The Embedded Patreon book club is reading Data-Driven Science and Engineering: Machine Learning, Dynamical Systems, and Control by Steven L. Brunton, J. Nathan Kutz. PDF book and links to lectures are at databookuw.com. Some recent links of interest: Datasaurus dozen: a collection of different small data sets that have the same summary statistics. You can see all the graphs here. From Humble Pi: A Comedy of Maths Errors.  The I Saved a PNG Image To A Bird YouTube video was really neat, which led to AudioMoth ultrasonic recorders as well as making a $40 bird identifier with an RPi and some software (BirdNET-Pi).  Chris is taking a course from Dogbotic. See their workshop list. We interviewed Kirk Pearson from Dogbotic on Episode 491. Transcript Mouser Electronics has a dedicated Empowering Innovation Together hub that covers the latest breakthroughs in tech. Their new series explores how AI is reshaping engineering—from design automation to rapid prototyping and predictive maintenance. You’ll find insightful articles, podcasts, and videos that showcase real-world applications across industries. If you’re ready to see how AI is powering the next generation of engineering, head over to Mouser.com/empowering-innovation.

Hello and welcome to Embedded. I am Elycio White here with Christopher White. This week, it's just us, so I'm going to get started right away. When we left Winnie the Pooh, I don't remember what he was doing, but Roo was in rabbit's possession, having been kidnapped. so that Kanga would leave the 100-acre wood. And Kanga had piglit, pretending she did not recognize him. She was making him do horrible things with the full confidence,

the faith that Christopher Robin would not allow anything bad to happen in his kingdom. Okay, are you ready? Sure. So I just go ahead and read. I guess so. Do you have other things you want to talk about? No, this is fine. Okay, then.

Oh, so I was thinking about starting a new podcast. Okay. I gave you some... Because we're doing so well at this one. I gave you some suggestions. Did you think about them? I didn't think about them, except I don't think we should do the podcast about pizza,

because I don't think that's more than three episodes. I think we could drag out pineapple or not for several episodes. I don't think interviewing... There should actually be a lightning around question. I don't think interviewing people's pets is really a good idea because it's hard enough to interview human guests

and have maintained conversation all the time. But the editing you did on that cat podcast was so good. I listened to it recently and I just once again basically fell over laughing. And nobody wants to hear a weekly compilation of my drum practice. Doesn't your drum teacher have to listen to that? Yeah, that's not a podcast. All right then.

Do you actually want to discuss this or is that a set up for a joke? I mean, we did talk about... You had Sleepy Time Science. Right, right. Okay, listeners. So I've done some research on this. So it's also, it's very difficult. so two things you discussed doing a completely separate podcast right which is fine that's easy um sleepy time

science where you would read science stories or something in a soothing sleepy time voice for able to nod off to because because what our listener said that his wife can't listen to the periodic table of elements yeah the show you might as well lean into that in the car because she falls asleep because I talk about nonsense and and my voice is soothing so let's just lean into it. that so if we get a thousand people to respond saying they want to do that we should definitely do that so write in people

listening I know there's more than a thousand of you so if you want sleepy time science let us know and I wouldn't read the periodic table I know no it would be good stuff I would modify things from from Wikipedia and talk about you know and there's plenty of open source

or public domain science books that are pretty good I've got one up there from the 1850s with a flower pressed in it? But yeah, it would be, you know, sciencey stuff, 20 minutes, me being soothing, plus, you know, five minutes of, of take a deep breath, feel your toes. There is some effort involved in producing a podcast, so I would want somebody to say, you know, not a thousand necessarily, but more than five people to say, yes, please do this.

And it would be separate. Yeah, it would be separate. And I don't think the Sleepy Time podcasts that have ads are just, I don't listen to them, they're too irritating. I don't think we'd have ads on a Sleepy Time podcast. But it would be Patreon and it would be like only one or two was available for public listening and you'd have to join to get access to all of them. So along those lines. So that's one idea. The other idea is if people are interested, you know, we've gone back to a bi-weekly schedule.

Yes. Other podcasts have memberships. Now, I know some people have support us on Patreon, which we really appreciate. Really appreciate. They don't really get anything extra except giving us, you know, their dollars. But what if? Well, they get the slack, although you only have to pay a buck to get the slack. You don't have to be an ongoing member.

There are some other podcasts that do subscription memberships, like Patreon, where if you give them your $5 a month, they give you some member special stuff. So they have a special podcast every month, and they have podcasts without ads. But the special podcasts are usually not topical. Right, the special podcasts are still. So for example, for our podcast, we might do a tier ranking of microcontrollers and have a guest to come on and the three of us would rank our favorite microcontrollers. And we'd have a set list that we were going to go through and then we'd all talk about them and rank them.

Right. That's totally copying this other podcast idea, but, you know, they didn't trade market. It's kind of topical. Didn't trade market. Movies, we could have a guest on. We could, well, watch a movie, not on the podcast, but watch it before the podcast, and then talk through it like, you know, something embedded related, you know, some sci-fi movie or sneakers, which we just watched, things like that. And then there's stuff that's way off the left field. But, you know, once a month, we might do something like that, have an extra guest.

and have a less, you know, an out-of-band podcast that does not, you know, part of the, it's not necessarily educational or a part of our normal interview schedule, just something fun that people who subscribed monthly would get every month. So that's another thing I'm trying to gauge interest on. I did mention the ads. A lot of those podcasts, other podcasts have, if you remember, you get an ad-free stream of the existing podcast. That's rather difficult to manage without doing everything through Patreon.

So Apple has a podcast subscription stuff, which we could do. But that requires everyone who wants who's interested in subscribing and getting the benefits to use Apple podcasts, which I don't want to make anybody do. I don't even do that. We can do it through Patreon, but that would mean you would have to get your your extra stuff ad-free or member specials through Patreon. But there's not a really essential way to like, okay, here's your special RSS feed person who, you know, pays us.

So I looked, I did some research, Libson, which is our podcasting host, used to have something like that, but it doesn't seem to exist anymore. And they push people toward the Apple subscriptions, which I don't think is a very viable option. So anyway, if you would be interested, if that's something that sounds good to you, let me know, and if you were willing to do it all through Patreon. We might be able to do Patreon and coffee somehow, but that would pretty much be it without building my own CMS. And that's what the people on the other podcasts do. They have their own CMS. They manage all the subscriptions themselves. They can generate private RSS feeds for every member, and that's just something I'm not doing.

No. Not doing. No. Yeah. So there's some stuff we're thinking about, you know, expanding the podcast a little bit and giving Patreon and coffee members something extra. But it would require a little bit of effort to get the extra things out of Patreon and Coffee. I think there's a way to get a Patreon RSS feed. Yes, definitely. So that might be something that people can do.

I know everybody's not in love with Patreon. Guess what? We don't have a better option. in love with anything on the internet so if somebody wants to write me a CMS and maintain it for free then then we can talk so yeah

does that cover all the stuff on the extra podcast okay okay I don't really know how it would work but just gauging interest yeah yeah since Chris isn't

doing as much technical work I figured I would just burdened him with more audio work. Do you like... Work? No. Do you enjoy the podcast process and editing and posting more than you enjoy doing contract work? No, but it's easy. It's like, you know, it's a...

It's not a... It's not a challenging thought process task. It's something I can do for a couple of hours that I know how to do, and it's always going to work. And when you're done, you're usually done. I'm done. Yeah. So it's, you know, it's like a household chore. Sometimes those are not enjoyable, but, you know, at its best, technical work is way more fun. But at its worst, it keeps you up at night.

Yeah. Before we get back to today's discussion, I want to share a new resource for anyone curious about the rise of AI in engineering. Mouser Electronics has a dedicated empowering innovation together hub that covers the latest breakthroughs in tech. Their new series explores how AI is reshaping engineering, from design automation to rapid prototyping and predictive maintenance. You'll find insightful articles, podcasts, and news. videos that showcase real-world applications across industries. If you're ready to see how AI is powering the next generation of engineering, head over to mauser.com slash empowering dash innovation. Now, let's get back to the show.

But you've been trying to do a lot outside of technology. Yeah. Well, and when we say technical, it's kind of weird because you're taking... It paid technical work. All right. I'll do that. So, yeah, I'm doing too much. So, okay, at the risk of this being a therapy session, I have competing problems with myself. One is that I don't like to work that much. There are times when I enjoy work, but it's, you know, it's kind of like the 15 minutes of excitement, followed, you know, and eight minutes of sheer boredom and stuff or frustration. Anyway, it's fine.

but if I'm not working if I'm not busy I have other problems like a little weird and so I've been trying to fight that while not having too much contract work by filling my day with more stuff and trying to learn more things

and so I've piled up way too many things well you had an intensive drum course but it's finished with that so now you have a medium intensive I'm sort of finished with that but I'm yeah so that's over but I need to actually do music now instead of practicing like crazy so I have a backlog of songs of songs to work on for

this record that someday will come out I added the dog dog botics course I took it the dogbotics drum machine course which I want to talk about at length separately from this topic so I'm doing some electronics I still have a client and they're doing field testing in the last two weeks I've been doing more stuff for them I'm trying to relearned French. I have a stack of French books, and I've been going through Duolingo, which isn't great, but it's keeping me, it's getting those cobwebs dust it off my French knowledge. I'm trying to read. I'm trying to keep up with the house. I'm trying to, yeah, just a lot of stuff. I'm forgetting a bunch of things. Like I, you know, I do want to read about, you know,

learn Python better and stuff. And I keep thinking, oh, well, maybe I'll write a video game or screw around with that. or anyway, it's, I've piled on too many things I'm trying to learn at once, and I've created a situation where I like, oh, if I'm not doing all those things every day or not keeping up, then I'm feeling like I'm behind on stuff that I'm just completely inventing for myself. So I'm trying to whittle it down to, let's, you know, let's try to do two or three things a week. That's the main focus, not eight. Yeah, because I've got other projects that are in the garage that I want to finish.

and I'm not playing guitar and it's like, there's too many things I want to learn. And it's hard because, you know, it's kind of a weird midlife crisis like thing. It's like, ah, you know, I'm over 50, and I haven't learned everything I want to learn. And I want to be good at everything, which is all stupid.

But long story short, I've added too many things to do and it's too much, it's too much for me, not from my I can't do it all perspective, but I can't do it all justice. And if I don't do it all justice, I feel like I'm failing somehow. So it's the 12 projects. There's, you know, one project or 12 projects thing, except it's not really projects. It's 12.

It's like I, you know, I want to go back to college and take 16 classes at once, which is, as the advisor kept telling me, you can't do that, please drop some classes. And I was like, but I'm only failing too. anyway so well and I should point out that many of these things are not projects learning French is not a project

right right no it's it's whatever and even music is not a project that's just something I do even your album it's it is a project but it's it's a super long project each song should be shorter and that doesn't count

the intense drum instruction you've been doing. Learning electronics, yes, the dogbotics

course is a series of projects, but learning electronics is not a project. Yeah.

It's a lifetime of wondering why you misplaced that wire again. Yeah. Well, yeah,

we can talk about that separately. So you've been on kind of a sabbatical, although you keep getting interrupted.

Okay. Yeah. And I was kind of hoping the sabbatical would be, you know, where do we want to take the podcast? And what do you want to do with your time? What makes you happy? Yeah. Because if you aren't happy, then you might as well be working because then you're happy and have money. The look I'm getting right now just indicates to me I have done this all horribly wrong.

That's okay. Continue. I don't know that I came into it with the right attitude. First of all, to stipulate, I don't think, I think I'm learning that my happiness is not necessarily 100% tied to whether or not I'm working. So I don't think, I think I would be more unhappy if I was working full time than I am right now, just to state the level set. Yes. Okay. Continue. But it is also interesting that you have put so much burden on yourself that in some ways, even without working full time, you aren't joyfully skipping through life.

I don't do that. That's not me. I cannot sit down and play No Man's Sky for eight hours as much as a day, as much as I want to do that because it's like eating pizza and ice cream for every meal. great until you do it. For me, like, it would be great for a couple days to do that. And at the third day, I'd be like, wow, I have done nothing. And, you know, this is something I'm discussing in other contexts, this feeling of needing to do things. And maybe that's, maybe that's my failing, but. Needing to do things, needing to learn things, or needing to accomplish them. Was there all subtly different aspects of the same thing? But, you know, I cast it as needing to learn things, but how do you decide if you've learned something? It's whether you can accomplish

something with it. You pass the final at the end of the class. Yeah, well, I don't have those for me anymore. I know. Isn't it weird that it's kind of sad that we miss finals? I mean, it's weird to miss finals. Miss finals. Oh, sorry, just me. I have, still have horrible memories of trudging up to the eighth floor of the library with my six-hour take-home final and sitting there and not moving for six hours while trying to answer, like, complex analysis questions. I don't want to do that again. Or, my favorite at the grad school finals, that started out as a 50-minute in-class, and then when it was clear that nobody in the class had gotten past the first of eight questions,

the professor says, well, this is now a 24-hour take-home. See you tomorrow. That's no good. Hard to judge how hard tests are, especially at that level. But anyway, you are going somewhere with this. There is seldom a I'm done. No, there shouldn't be. So one of the things talking to Dimitri recently was nice

is that he had I'm done posts. Yeah. And I so seldom actually have those. Even with clients, it's not always a, okay, we've shipped it. There's always a little bit of trickle. No, the only done I ever feel. with companies or clients is when I quit. No, and that's not a joke.

Like, yep, I'm done. This project is done because they're not paying me anymore and I don't talk to them. That's it. I don't have to think about it anymore. Any other situation is there's always something that could happen to your product that the thing you've worked on

that could require you to work on it, right? And that's always true. You know, full-time companies. There's like, well, something could still go wrong. We shipped that, but now I'm in support mode. Okay, I'm in support mode, and I'm working on the next thing. okay, I'm supporting the next thing, and, you know, it's that roller coaster.

I've never, you know, the reward for a job well done is always another job. I don't know. I am, you know, slowly edging into something that looks like retirement, and I don't know what that looks like for me. So I'm trying to figure that out. But we can talk about, are you trying to help me, solve my problem? No, I mean, explore it. Yeah.

Yeah, I don't think there's a solve. I don't know that there's a problem. I mean, one thing I've been trying to do, which is, I think to some extent my brain is addicted to information. And it doesn't need to know, it doesn't matter what information it is. I can fill it with garbage. I can fill it with, you know, reading a book. I can fill it with working on a project or doing music or stuff. but, you know, there always seems to be some,

I need some kind of input into my eyes or my ears or something, you know, go for a walk, I'll put a podcast on. And so what I've been trying to do in the last few weeks is not do that for at least a few minutes a day and have zero come into my head. And doing that is very surprising because it just reminds me how often I have something in front of my face. A book or a game or something in my ear.

years or something I'm doing and I so rarely give my brain a break to not be shouted at and it was really good because a few of the times thoughts started coming into my head those things that come out of your brain

with words and pictures that aren't coming from the outside it was weird so you know musical ideas just came up after I sat and didn't do anything for a while. This isn't meditation necessarily. It could be meditation, but those are distracting during meditation.

But just sitting outside and staring at the trees for 15 minutes with nothing near me that is giving me any, you know, words or stuff. I think I need to give my brain a break more often than I do, a real break. Not so that it's not doing anything so that I can actually see what. all the stuff I've been putting into it maybe comes together into. I think, and don't take this as a... I already have. Right, okay.

I think daydreaming is super important. What do you mean you don't daydream all the time? Do you? Oh, yeah. Okay. I mean, sometimes it's outside and clouds always play a part for me if I'm outside. But being at the beach without the little dog is usually some element of daydream. Yeah, but how often do you not have origami or a book or a podcast?

I think more often than you think. All right. Because I'll have my headphones and my ears, but I won't be listening to them. I do often have a book. But sometimes I'll have a book and I'll, you know, unfocus and daydream. Anyway, I'm trying to experiment with doing nothing. Daydreamers Anonymous. No, it's not even daydreaming necessarily because, yeah, I mean, it gets daydreaming, but that makes it sound.

No, that's good. That's a good word for it. I mean, that's totally fine. No, I mean, but there are probably more official sounding. No, it doesn't need to be still strongly. Yeah, that's fine. Less, obviously, I'm intentionally floating. Just letting your brain do its thing without trying to stuff more into it.

Yeah, which if daydreaming is what's happening, then that's good. So, the title of this section was burnout. I thought I was doing better until some work came up. It was some support work, remote support work. And it wasn't actually a lot. It wasn't a lot, but it was stressful because the stuff I had worked on was not working. And it's very difficult to support. And so it wasn't so much that it was a lot of work.

It was just I got super stressed about it. And I'm still kind of stressed about it. And there was a time aspect. People were waiting on you. There's still a time aspect. This is ongoing. It's not over. Well, your part is for now until they tell you the next bit.

Yeah. Yeah. So I dealt with that more poorly than it's not, I'm overworked. You know, it's not, oh, I had to work a lot. It's not, oh, I had to work two hours and what a baby. It's, I just am having still emotional reactions to work and connection to projects that is not super healthy. Whereas I took on more work, even though I was fully loaded. I am an idiot. I'm a past client came and needed some help and their time frame is pretty big so I'm squishing in what I can but I visited the client site and remembered why I love this client

it's just I go there and I didn't participate in the lunchtime conversations because I was off to my side and what I was eating was, I didn't really have lunch. I had a snack that I was pretending to eat was lunch. And I didn't want anybody to comment on my food. I was being weird. I was being very awkward.

And yet I was eavesdropping. And there were some people, and they were talking about camera traps, and I wanted to go and talk to them and talk about what they knew and what I'd learned from Meredith and Akiba. And then they were talking to some about sampling, and I wanted to go over and talk about this compressed sensing that I'd been reading about. And I just wanted, and then that group left. Another one came in and they were talking about science communications.

And I wanted to ask so many questions. And man, I just nerded out for like the half an hour I was there to eat my lunch. And it was so fun. And I mean, the rest of the day was actually really fun too. I don't get to spend a lot of time on this client. but I'm kind of glad that as long as I'm going to accept a little bit of overwork, then I'm going to be doing it with people I really like. So that was fun.

I didn't really mean to take on more work, but I don't have any mentoring clients right now. So that's probably going to fill things out. But I'm also trying to make sure that I don't work too much. Definitely the number of steps I got yesterday as I was mind-melded with my computer indicated I needed to move around a bit more. But then after I worked, I conned Christopher into going to the beach for a walk. And I think he thought we were just going to go and, like, take a short walk.

But I took the dog, the dog and I went off and played and ran around and generally got our fidgets out. And Chris was just there in shorts and a t-shirt shivering. It was very sad. It was cold. It was like 8 o'clock. Did you know where we were going when we left the house? It was 80 in the house when we left, so I wasn't aware that the beach would be cold. Really?

Yeah, really. Anyway. Do you have a jacket? I had the emergency puffy jacket. So embarrassing to wear the puffy jacket. So, yeah, working too much, which is why I have not read the chapters for the technical book club. But I have to do that really soon. I am enjoying the technical book club

Do you want to talk about that? Yeah. On Patreon, we have the book club channel. We are reading data-driven science and engineering, colon machine learning, dynamical systems, and control by Stephen Brunton and Nathan Kutz. And we are, oh, I think we're in chapter four. We're about to finish chapter four.

Yeah. Regression and model selection. It's been interesting. There's been a couple of mind-bending pieces that I didn't really expect. Thinking of control systems, PID loops, motor control, as data problems. Like neural nets are a data problem. What is a data problem? You have a bunch of things.

you have a bunch of measurements which represent a system that you don't really know all of the system. How do you make it so that in the future you can make determinations about the system you don't know all of and based on the measurements that you are reading? So like in machine,

learning, that would be you're going to get pictures and you have a system that ideally tells you what the pictures are that maps actual images to names, but you give it this new data, which hopefully will, if it's a tree frog, will tell you it's a tree frog. But you also know that there's imperfections in that. and in doing a motor control, you can't model all the friction. You can't model all the damping, you can't model, you can model the static friction, but it's hard in your equations. Anyway, I really like that aspect of it.

It's one of those things that just keeps coming up. But isn't that, not to just be a little it or dismiss it, but isn't that partly because everything boils down to linear algebra at some point? Yes, yes, that is true. It is a book that is a lot of linear algebra. In fact, one of the recent sections heads, I'm going to show you six different ways to solve AX equal B in Matlab. And it's like, okay, is that really necessary?

But it turns out that there are different reasons for solving different things. So this has been going on a while and you've had a good group? Yeah. I've had a dozen. Few people have wandered off for vacation. But I have a couple who say they're coming on for one of the future chapters. And we're doing about a half a chapter a week, which is about 15 to 20 pages. So what will you be able to do with having gone through this book?

What is the average person in the book club looking to get out of it? Or what are you looking to get out of it? Okay, so Tom mentioned the weekly book club meetups as Fancy Math Anonymous, which I thought was pretty funny. Because this is the fancy math that I was talking about before. This is being more comfortable in approaching things in an analytic method. And one of the things that I have gotten out of this is the next time somebody hands me a machine learning regression model, I have the words to say, it's really dumb that you have two of these things that are essentially the same and they have different weights. It should choose one of them and use that.

it shouldn't randomly choose all, and I have the words to talk about how important interpretability, how important it is to interpret the models, that they shouldn't be black boxes and that the regressions that give you a black box full of terms are usually provably not as good as a model with fewer terms that are more specific. And I can suggest algorithms that will help make them more interpretable, make them simpler. Okay. Now, is this mostly control systems focused? You mentioned PID, and I think there were some other things like that.

Okay, because that's what you're working on for work. Were these two connected in any way? Well, this is how the control systems and the data part are somehow interlinked in ways I never consider. It is a lot of linear algebra, and it's linear algebra in a way I didn't expect. Okay, so you have X, Y, Z as a coordinate system. And we've talked about normal vectors. If something's in X, Y, Z, and Z are the normal vectors, the unit vectors. Unit vectors.

They are normal to each other. And they are normal to each other. And so if you want to put something in space, you could put it in X, Y, Z. And then maybe you, that's in your person frame. And you have a car frame, and you can translate between the car frame and your person frame with a rotation matrix. You've already lost half the audience. Okay. Anyway, you can move things around in space, and you can describe how they're moved. And that form of linear algebra is fine with me.

Okay. What if, and hear me out here, instead of X, Y, Z, we use Fourier. We use Fourier as unit vectors, like 1 hertz, 2 hertz, 3 hertz, 4 hertz, as unit vectors. And we use DFTs, we use Fast Fourier Transforms, to put our image or data or location into Fourier space instead of, of 3D space and treating Fourier as just another basis vector. I tried to explain to you how cool this was 21 years ago, but now you finally agree. And I finally agree. Spectral methods for solving DEs.

Yeah, we went through spectral methods for solving DEs, but also the idea of all of these different basis vectors. Yep, yep. And there's not just the Fourier ones. There's, you can develop offshoots of four-year-ones. You can invent your own. You can adventure as long as they're, or, you know, as long as they form a orthornormal basis. And so the book actually is, is headed towards, I mean, we started with SVD.

Singular value decomposition. Which is basically inventing your own basis, inventing your own axes. Right. based on your system. Based on your data. Your data, yeah. And then you can, you can now put new data in the basis of the old, just like you could put a position in the car's frame versus your own frame.

Yep. And switching between frames and all of that. And it's, this is, again, we're switching between, a bunch of tools that I knew for data analysis. And we're applying them to complex systems. I tricked you into liking physics. We're looking at data analysis, which I already knew was related to machine learning. That I understood, the regressions, all that makes sense as far as machine learning.

But now we're also applying them to this other systems thinking, which is just bonkers to me. Oh, well, you talk about the bird. a PNG thing? Which is, this is a separate video that Chris brought up and it turned up. I feel like you were going to hand me a dollar. Okay, go. And then it turned out that it had a whole bunch of stuff

that we talked about recently. There's a YouTube video. I don't remember where I originally saw it. Maybe YouTube recommendations. But the YouTube thumbnail and the headline is I uploaded a PNG to a bird. And it's very silly headline. and thumbnail, of course.

It looks pretty bad. It's actually a really good video from somebody who does, I don't know what they do. They do a lot of field audio stuff, it seemed like. That's their channel and a lot with wildlife. And they talked about, mostly the video was talking about recording birds and looking at their bird song on spectrographs and some of the anatomy of birds

and how they can produce multiple fundamental frequencies at once, not just harmonics. I mean, humans can sort of make more than one tone at once, but we only have the... It's all based on harmonics. That's normal voices are based on harmonics. Yeah, normal voices are based on harmonics. But birds have a couple of anatomical features where they can produce two or three or four sound. Chords.

If you like, yes, chords. So he was looking at that, and he was talking about how they can be, they go up to ultrasound. So if you want to do real field work recording birds, you should get... There was some technical discussion of the equipment and stuff that you can get. But anyway, he had this idea, since he had all this stuff, of playing back, he met up with this person who does animal rescue, and they had a Starling. A Starling, they're very, learn all of the audio. Starlings like to learn and mimic sounds and things.

He put just a kind of a stick figure drawing of a bird and a spectrograph. And then he played that. And it sounded junkie. I mean, it was, you know, basically since it was two lines, it was two fundamentals that were sort of warbling as the outline of the bird continued, or whatever. But it was two fundamentals. Yeah, because he had the bottom of the bird and the top of the bird. And I think maybe there were some other parts. There were a couple other features.

Long story short, he plays this to the starling, not expecting anything to happen. And nothing does happen for a while. I mean, he wants something to happen. He wants something to happen, but it plays it to the Starling, hoping it would mimic it and play it back so that he could see the bird in the Starling's spectrograph. And I guess he decided it wasn't happening. They recorded a bunch of stuff with the Starling after doing this and didn't sound like it was happening. But as he was reviewing it, there it was. The bird was the picture of the bird was in the recording of the Starling.

So it had played it back. He just hadn't identified it at the time. And he did some other very specious data analysis of how many bytes he uploaded to the Starling, which just, no. But it was pretty interesting. It was kind of a strange idea, and it actually worked, and, you know, I suppose you could, you know, put secrets into Starling and some sort of encoding, if you wanted. It's hard to get it out on demand, I think. But so that leads into something. It was just a really good video that had a lot of different techniques put into it

and it wasn't like you needed to understand the math and all of that because he did a good job of hand-waving over it and yet it was really deeply technically interesting the methods he used and he went through all of the hardware and what he liked and what he didn't like And for 30 minutes, it was packed full of pretty good information from spectrograms to frequencies to starlings to audio moth and bird pie, which are bird net pie. And Merlin. Merlin and databases. You know, they talked about how to set up a permanent field recorder in your yard to just give you a listing of every bird adheres all the time.

And audio cameras, which are cameras that then find where the audio is coming from, although that was a more expensive and difficult proposition. So that somehow made it back into the book stuff because it was relevant and cool. One of the other things that went into the book stuff from Ristow, one of the people there, he found in a book A Humble Pie, a P.I. A comedy of math errs. A resource called the

Datasaurus dozen. Mm-hmm. Datasaurus doesn't. Datasaurus doesn't. Like a dinosaur. Like a dinosaur. Uh-huh. And it's a dozen, obviously,

small data sets, really small. 40 bytes each 40 samples each but these different data sets if you plot the X and Y they have different images one of them is a dinosaur one of them is a star one of them's circle

one of them's two circles one of them's two parallel lines blah blah blah all these different patterns that you see when you plot X Y but they all have the same means for X, same means for Y, same standard deviation for X, same standard deviation for Y, same correlation between X and Y. Now, these are our basic stat set. This is the stat set you start with. And yet, they have these dozen sets of data that are obviously quite different, but have the same essential stats. I mean, I think, you know, if you said, give me the essential stats, these would be the ones, the means, and the standard deviations. And yet they lead to such different data, showing us that maybe our stats are limiting and that plotting the data is always the right answer. It's hard to see things as visual creatures, you know, without looking at things from different perspectives, whether it's transmitting it to a plot or.

even making things audible like birds or making things that are audible visual like spectrographs that's you know as a reason we do all those things it's because our brains are funny we're designed to we have visual creatures look for food and predators so we have to turn our data into predators and that's how you ended up with the data source uh the history of the Datasaurus doesn't was kind of cool too. There's a set of four with some name that starts with A which I will put in the show notes.

But these dozen were found with simulated kneeling starting with the dinosaur and trying to find other things that could that had the same statistics but were

very different shapes. They aren't It's just amazing to me that they aren't. I understand you could get a dinosaur and then a random set of data to match. But then you have the star and the circle, and it's just, it's kind of bonkers. Let's see. One other thing that we did in Book Club was looking at mathematical functions. And this was again from Tom.

He has a bunch of books. physical books, although somewhere online as well, that plot out functions. Like, I mean, okay, I kind of know, I know what a parabola is. Paraly. Go to me. And I know that there's an equation that makes a heart because I'm pretty sure that was one of EMSL's Valentine's. The cardioid.

Cardioid, sure. there are a few other functions that I know they exist but I actually don't know what they are but he had this books multiple books multiple big books of curves and surfaces and the equations used to generate them and I have to admit I was looking at them like going I wonder if that could be applied to origami

because I get a little bored of some of my wave patterns and I'm ready to level up on that, but my Python script is a little out of date right now. Anyway, anybody wants to change my Python script so that it outputs Bezier curves instead of points. I'm all in. You should be able to do that now. You're doing all his fancy math.

Yeah, but every time I try to do it with one of the AI helpers, it goes very badly. Huh. so do you are you enjoying the book club do you think you will do another book should we should we say if you're interested in doing another book uh join us and discuss what book to do next well i mean we're we're in chapter four so so it'll be 2017 2017 wow what year do i think it is i don't know if who cares uh 2029 um maybe 2026 okay all right uh But yeah, Phil Coopman has a new book coming out about AI and safety. This one's been really good because it was stuff, like going through the table of contents, it was mostly stuff I knew how to use. And so even if I don't follow all the math, I'm not really falling behind because I know the next chapter a little bit.

You know, the outlines of where this is headed. And that's been really good. And then the support for this book with the videos and the website have just been fantastic. So I don't know if I would be able to find a book as good as this. I would like to know more deep Python. Yeah, you could go through Nanda Tetris with a bunch of people. Or something along those lines. That one doesn't appear.

I mean, I went all the way through Nant to Tetri so that one doesn't appeal to me. But yeah And right now we have one book going But there's no reason why we can't have multiple books going If people want different ones Well, unless you want to run every Oh no Somebody else has to be the lead on some other book

But I'm enjoying this book I would consider being lead on another book club choice But I would also encourage someone else to I mean it doesn't take that much more time being the lead the hard part is committing to actually doing it. And then coming up with jokes when nobody wants to say anything. Okay, sorry, that was really long. Do you want to talk about your dog-dog-dog-bottic course?

Sure. Do you think we have time? So, yeah, as mentioned in my discussion of my continuing something, I'm taking this dog-botics course. It's, we had Kirk Pearson on the show a few months back, and they're the founder of Dogbotic, and this is a place where we've mentioned before, has various workshops and things for electronics and visual media and the audio media and other things you can do. They're about eight weeks, a couple hours a week, and they send you all the materials, and it's pretty cool. And so I decided to take one of these after getting re-energized by Kirk's episode.

and getting their book. And yeah, it's not an ad. They didn't pay me. I didn't even use our coupon to get a discount, so I just paid for this myself. But I've been really enjoying it. So the one I'm taking is called Do Yourself Rhythm Widgets, I think. It's basically making an analog drum machine from discrete parts.

Well, and some ICs. But Logic ICs, you know, CMOS. For a whole ICs. CMOS, Logic ICs, Inverters. logic gates, phase lock loops, and some op-amps and things. So this is all breadboard-based stuff, and we're building, you know, oscillators to make different sounds. We're building different triggering systems to make the sequence happen, and making a little

mixer with an op-amp to put all the sounds together so they can be adjust the volume. So we're doing all that. It's very fun. It's very hard. Some of the circuits are, I don't have a lot of experience doing slightly more complex circuits on breadboards, going from schematic to breadboards. So some of the op-amp circuits have multiple low and high-pass filters and things that are big, not big, but they're networks of capacitors and resistors that are all going to different to the same points. And it's difficult for me to translate that to the Tetris that needs to happen on a breadboard. The electronics of it.

Yeah. not really learning the deep electronics. So, like, I'm learning, you know, modular circuits that I can put together to do things. So I know how to make an oscillator with an inverter, and I kind of understand how that works with the electronics. But the oscillator we make out of an op-amp that has two low-and-high-pass filters, I don't really understand what's going on there with the feedback and the op-amps. And I kind of have a vague idea of what the filters are doing, but I don't really. have a deep understanding of the electronics there.

And I still have some big holes in my electronics knowledge for, you know, how things work and how current flows and things like that. But I'm getting experience with it. I'm building circuits. They're working. It's fun. It's giving me directions to go look up other stuff. Like, oh, now I know I need to learn more about op amps.

And here's some things to look up. Like, how does a twin T oscillator work? And that will take me to, okay, that's too complicated. Well, how do it was a simpler oscillator with an op-amp work, and then I can learn that and move up. Gives you curiosity. Yeah, and playing with a simulator a little bit on Falstad to try to try to build some of these circuits and see how they work.

Well, that's not working that great on some of them. But, so it's been fun. I highly recommend them. They're about $600 for eight weeks of instruction, and that includes all the parts, which was a lot of parts. Like, it was a... The box was pretty big. Yeah, I mean, I got tons of capacitors, resistors,

bunch of ICs, three breadboards, three small breadboards that you can gang together. So everything you really need to do all the projects for the course, at least on a breadboard, plus a whole bunch of extra parts for experimentation. And it's two hours a week. Yeah. But you also have done some extra time outside rebuilding things when you weren't happy with them. Yeah, well, the drum machine that we're building got so complicated on the breadboards that I actually bought a bigger breadboard and then started.

rebuilding it because I ran out of space and everything was I had flying wires everywhere jumpers big ones do um and so I got more I got some of those variable length breadboard wires that are straight um that cleaned things up a lot but I had to basically I just okay some of this works but I can't make it work the way it is now because I can't even see under the wires it was such a rat's nest so breadboards are hard but yeah I dismantle all that and rebuild it so you do have spent some time out of class if you want to get the most out of it um although it is a build along course. He built circuits and you can do it in real time and he explains as he goes. So a lot of the first, at least for the first few weeks, it was like, yep, I'm building this as we go and it works

in class. So that was cool. But they have a lot of their courses. I'm probably going to take some more. And I have enough knowledge now that I can take this drum machine and actually build a drum machine out of it. I'm going to probably, you know, either. Replace all those parts with an Arduino? Well, not all of them. No. Well, it's not an Arduino. So the triggering stuff is really cool and complicated that we're doing. It's, you generate an LFO from an inverter, and then that goes to... A low frequency oscillator. Yes, a low frequency, a square wave, we had a certain frequency.

And this is generating a square wave that triggers the drum sounds. So, you know, it's at, you know... Some number of beats per minute. Four hertz or something, you know, right? Yeah, so boom, boom, boom, or whatever. So those, that, that, that, you generate... a low-frequency oscillator, and then that goes into a frequency divider, either a binary or a decade, and then you can choose which outputs, and that'll generate different frequencies

for the different drum sounds, and then those triggers go out to the oscillators to make them go at different rates, trigger at different rates, not play different frequency sounds. So, like, the bass drum might be every fourth beat, the hi-hat might be every beat, the tom or snare might be every second p. So you can arrange your clock tree to do that. It's very much a clock tree. And that's all fun. And yesterday we talked about using phase lock loops

to do similar thing where you can, instead of frequency dividing, you can frequency multiply, which is really cool, because now you can get subdivisions of the main beat and do fast things. That was weird that frequency dividing and frequency multiplying kind of mean the same thing.

Flammable or inflammable sort of thing. It's like wavelength versus frequency, right? Once, yeah. And that's all fun, and I enjoyed learning it. However, so there's one path you can go with that. You can make your drum machine, so you have all this clock tree, and then you either use rotary encoder, not rotary encode, rotary switch to adjust which drum gets which division or a patch panel, so you're patching the oscillator triggers to the outputs of

the dividers that you want, and so you can change the pattern with a patch panel of, you know, wires. But your patch panel right now is a breadboard with wires. Which is horrible, but stipulated I'm getting off the breadboard. I mean, you're going to solder it to Perf board. Or if I'm really ambitious, try to do something on KECAD to make a bad PCB for through whole parts. Don't email me. Or just email him the solution.

So anyway, but I'll get away from Perf board. So the question is, do I want to do it that way, which is interesting because it's all, you know, discrete parts. and there's some error involved with the triggering and stuff so it makes it more human. Or I can throw all that triggering stuff out and stick a micro on there and a bunch of DACs, or a, yeah, a bunch of DAX

and trigger that way and have more flexibility in a digital triggering to an analog drum machine. So that's probably the way I'll go, but I haven't really decided. And there's a lot more I could do with the micro, Like if I'm using a PLL, I could use the micro to output of voltage, which is just a clock and do some stuff like that. Or I could convert some of my drum sounds to have voltage controlled oscillators, which would change their sound frequency based on a voltage.

But anyway, it's cool. And I'm trying to give myself permission to not fully understand everything, which is the... That's hard for you. Really hard for me. Because if I do that, I'm not going to have fun because I will spend... six months learning how an op-amp works and then i'll be like well that was boring when i could have just said you know these are the standard circuits people use and there's a lot of ways to tweak them to make your own cool sounds and stuff let's focus on that instead of the details of you know and

i know how to calculate you know given the twin t oscillator and the parts what notch frequency band that that's going to correspond to but i don't really understand how the, you know, how the loops are all working. So stuff I do understand, but, and I don't know, they do, just, do that ready doing electrical engineering, know at a fundamental level how those sub-assemblies work? I would hope so, but maybe not. Like, maybe I'm being too hard on my electronics knowledge

by trying to understand everything at a, not quite a physics level, but... It's been so long since I took electronics. I wouldn't remember any of it. I don't remember any of it. And analog stuff was always kind of... It's been a good course, and I've enjoyed it. In your drum course, your... My actual drums course.

Intensive drum... Yeah. Yeah. You talked about metric modulation. Could you do that on your... Well, with great difficulty involving clock trees and things like that. And, yeah, I mean, I don't know how I would do it. But you can do polyrhythms pretty easy.

which is a similar concept where you have one meter playing against another meter. I just want you to turn the drum machine into your drum course. Well, that's some other fun stuff I can do because the one thing we learned as an aside was how to use Piazo-Paiso. I don't know how to say that. I never know how to say it. triggers, which are the little flat disks with a crystal in them. And when you touch them or hit them, they generate a small current. What?

I just like the way you're hitting your hand, as though everybody can see that. Anyway. And you're making this little circle. So we learned how to take one of those triggers, which is, and put it through an op-amp to generate a trigger signal that you could use to trigger the oscillator. which is exactly how electronic drums work. They have with slightly greater sophistication, Piazo sensors in the drum heads or near the drum heads

that when you hit it with a stick, generates a current or voltage, I can't remember if they're, whatever, generates electricity that's detectable by an amplifier that can go to a module that decides whether or not to make the sound. And so I could take my electronic drums and plug them into my drum machine and trigger the sounds with my drums and sticks

and things, which is something I plan to try to figure out how to do. Or you could take a micropython at a MIDI library. Well, yeah, but that's back to the sequencing I was talking about. Well, cool. I mean, that sounds interesting. It's been fun.

I keep looking at their, they have one about textiles. Mm-hmm. But then they say something about being able to knit. And I'm like, yeah. That's a prereck. Yeah, most of them don't have super strong prerex.

I don't know that it was a prereq, but I think there were like five things I didn't know how to do that you would know how to do by the end of the course. But they all seemed like five things that each took six weeks to learn. So I felt like I would go in without knowing any of the, I don't know. It's kind of the opposite of the data-driven books where I technically know this information. You're just reconfiguring it into a better system. I think I talked about journaling the last time.

It was just us. I'll be happy to know that I have fallen off the journaling wagon. Why would they be? Despite the sticker addiction. You had a pretty good habit going. What happened? You've written it all? No, there were two things that happened.

One, we had friends come and stay, and I didn't want to spend the time away from them. and then since they have left, I haven't started again. The other part is that I got stuck on a particular journaling prompt and not being able to answer it, I feel like I can never go back to my journal. Oh, that's okay. It's kind of sad, yes. And I tried to get you to answer the journaling prompt,

and you didn't either. So I couldn't even, every once in a while when I journal, I would ask Chris the journaling question, and then I would journal about his answer. Yeah, that's my fault. So that I didn't have to answer. Yeah, okay. No, now I understand how this works.

And so, I mean, there's a ton of stickers I haven't gotten. We've been seeing whales at the beach, so not only would I be getting beach stickers, I'd be getting whale stickers. Anyway, are you ready to answer that question now so I can go back to journaling? Go for it. Okay, you have to come up with three to five role models.

All right. Have you done that? Let's go with no. Yep, that's where I stop, too. Okay, so... All right, yeah. So, I mean, what's a role model? Is it somebody in your field who you admire their accomplishments?

Is it somebody who you want to act like in your daily life? Is it... Someone who inspires you to be a better person? I mean, because the danger of picking role models just out of the randos that exist in the world is that they turn out to be terrible people. And then people are like, why are you using that person as your role model? Okay, so the example was Mother Teresa Because there are following questions

That just make this harder and harder And then so the next question is Name a particular thing you Respect that inspires you about this person So like for Mother Teresa it may be compassion And then the next part Don't email us see this is very long

and the next part is to think about the opposite of that thing that you just said so for compassion it may be selfishness and then the final and this final part think about what the good things are in that question in that characteristic so for selfishness you might say it is important to understand your own needs and to take care of them as well as being compassionate about other people Okay, so

Yeah, I get to role models and I found a few Maybe just skip this one I don't know why you have any role models? Come on! I mean I, but like I said, they're, you know, they're more like avatars of how I want to play drums or... Yeah, Taylor Swift is one of mine too. I don't know that she plays drums very well. It doesn't matter. She does everything really well.

And, you know, yeah. it's probably that you're overthinking the question and you can just list some people who you admire for reasons that you want to aspire to that they don't have to be perfect people or encompass everything I guess but I can list probably a dozen musicians well okay not all musicians then

but go ahead go ahead go ahead come on I mean Gavin Harrison Nate Smith Vinnie Culliata Neil Pyrt Okay, Neil Pyrt, let's talk since he's the Rush drummer and I know who that is. And he didn't quite make my list

but he was in the top 20. What's a characteristic you like about him or find inspiring about him or is the reason he's on your list? Well, I mean, he was a very good drummer at the time that he was a professional drummer. Things have changed in drumming. 14-year-olds can play like Neil Pyritt now because of YouTube.

It's just bonkers. Anyway, but he was very innovative in what he did. He was the first person. He was in rock, you know, a person who took the instrument very seriously and thought about it analytically and thought about what he wanted to play and how he wanted to approach it. He often took his playing and deconstructed it completely. So he was very thoughtful. And went back and relearned things with different techniques. technique. He was very thoughtful. He was a writer. He was good at writing and he, you know, had a very

literary background. And very lyrical writing. He, right, and he wrote the lyrics for the band, but he also. Oh, that was what I meant. I just meant his writing's very pretty. But right, but he wrote books also. Yes. But yeah, and he was just sort of an interesting guy from a musician perspective because he was very thoughtful, he was innovative, he had very, very high caliber professional technique. He was also very resilient. His life fell apart. And he did what he wanted to. He did what he wanted to.

Yeah, he was very resilient. He lived through several personal tragedies. And he stopped playing when he decided he was done, which was interesting. And there were things he didn't enjoy, and yet he found joy in the things that he did enjoy and managed to continue through that. And unfortunately, he didn't live long past his retirement, which was not connected to his retirement. But, yeah, I mean, he kind of... Sum that all up in one word. See, I can't do this question.

Why would you sum things up into one word? That's why we have more than one word. Because then you have to find the opposite word. All right. The opposite is Lars Ulrich. I'm going to get so many emails. I don't know who that is. He's the Metallica drummer. Is this the one that said, I am the most humble person ever? No, that's their guitar player.

Okay. Anyway, Lars is great, but he's, yeah. But he's not a bad person. I don't know what the opposite would be. I don't like this prompt, and I think you should ditch it and do something else. And go back to just giving myself stickers. You always used to skip prompts you didn't like. I don't know why you didn't.

I just got stuck on this one. Yeah, all right. No, usually I make fun of the prompts that I don't like. If you're a Lars Ulrich, you can email me your hate. If you're anyone else, I don't want to hear about it. All right. And that's it for the show. If you want more shows, let us know.

Because you can have more of this. for a low low price of $5 a month, let's say. We do have some good guests coming up. A couple of authors, a couple of make types. Lars Ulrich? Yes, Lars will be on the show to defend his honor. Just wanted to practice. Ouch.

Wow. Wow. Thank you. for listening, thank you to our Patreon subscribers for giving us questions, which they did not actually do this time. They did. They were in the middle of it by the time they arrived. Oh, how does one start to do things in a self-motivated way?

I hear Ritalin and other ADHD drugs are good for that. Otherwise, I have no answer. we can take that up as a real question next time yeah there is no shortcut sadly sorry bribes if you can bribe yourself or have somebody else bribe you stickers it's funny i was thinking recently about how many people are far more accepting of lots of different things which i think it's great i think being accepting is good uh but that there have been times in my life where I needed the structure to force myself. Yes.

And then once I get past that foreseen, I find the fun. But, I mean, engineering education is not something, there are days when you don't want to continue. And so I don't know how to balance the... It's accountability. for me motivation is something that requires accountability i need somebody something or someone external either keeping tabs on what i'm doing or encouraging me or getting me over the hump when i'm stuck on something i have a lot of accountability and that's why instruction is good that's why schools

exist because most people are not great at being autodidacts unless you know they're really into stuff Yeah, for most of the things I've done, you know, jobs are in accountability. I think most people, given the choice between doing whatever they want and working for a company, you know, doing whatever you want and having everything provided for and working for a company every day, a lot of people would choose the former, I think. And so, yeah, that's what the money is for. That's what the money is for. Anyway, thank you to our Patreon subscribers.

Thank you to your glasses. We're falling. Thank you for the dog. So much gratitude here. We do appreciate you listening. And thank you to Mouser for... You've paid it this far. For sponsoring the show.

Let's see. I guess that's it. Oh, if you have any comments, questions, or are Lars, you can contact us at show at Embedded.fm or hit the contact link on Embedded.fm. I will have show notes with whatever links I remember, which probably won't be all of them, so feel free to contact me and make me put up some more links. And now, as I mentioned, Rue has been kidnapped, and Kanga knows that Piglet is not Rue, but is very calm about. that. Before he knew where he was, Piglet was in the bath, and Kanga was scrubbing him firmly

with a large leathery flannel. Ow! cried Piglet. Let me out. I'm Piglet. Don't open your mouth, dear, or the soap will go in, said Kanga. There, what did I tell you? You did that in progress! Sputtered Piglet as soon as he could speak again, and then accidentally had another mouthful of lathery flannel. That's right, dear. Don't say anything, said Kanka. And in another minute, Piglet was out of the bath and being rubbed to dry with a towel. Now, said Kanka, there's your medicine and then bed. What medicine? said Piglet. To make you grow big and strong, dear. You don't want to grow up small and weak like Piglet, do you? Well, that. At that moment, there was a knock at the door. Come in, said Kanga, and in.

came Christopher Robin. Christopher Robin, Christopher Robin, cried Piglet. Tell Kanga who I am. She keeps saying I'm Rue and I'm not Roo, am I? Christopher Robin looked at him very carefully and shook his head. You can't be Rue, he said, because I've just seen Rue playing at Rabbit's House. Well, said Kanga, fancy that. Fancy my making a mistake like that.

There you are. said, Picklet? I told you. I'm Piglet. Christopher Robin shook his head again. Oh, you're not Piglet, he said. I know Piglet well, and he's quite a different color. Piglet began to say this was because he had had a bath, and then he thought perhaps he shouldn't say anything at all. And he opened his mouth to say something else, and Kanga slipped the medicine spoon in and then patted him on the back and told him that he was really quite nice taste when you got used to it. I knew it wasn't Piglet, said Kanga. I wonder. who it could be. Perhaps it's some relation of poos, said Christopher Robin. What about a nephew or an

uncle or something? Kanga agreed that this was probably what it was and said they would have to call it by some name. I shall call it poodle, said Christopher Robin, Henry Poodle for short. And then just when it was decided, Henry Poodle wriggled out of Kanga's arms and jumped to the ground to his great joy Christopher Robin had left the door open. Never had Henry Poodle Piglet run so fast as he ran then, and he didn't stop running until he had quite gotten to his house. But then, when he was 100 yards away, he stopped running and rolled the rest of the way home so he could get his nice, comfortable collar back again. So Kanga and Rue stayed in the forest. And every Tuesday, Roo spent the day with his great friend Rabbit. And every Tuesday, Kanga spent the day with her

great friend, Poo, teaching him to jump. And every Tuesday, Rue spent the day with his great friend, Poo. And every Tuesday, Piglet spent the day with his great friend Christopher Robin, so they were all happy again.