• 8 months ago
Today on the flagship podcast of canceled-out cricket sounds: David, Andru, and Will react to prototype headphones that use AI to take noise cancellation to a whole new level. Michael Fisher, aka Mr. Mobile, joins the show to discuss mobile phones with physical keyboards and his latest project, “Clicks,” an iPhone case featuring an integrated keyboard.

Category

🤖
Tech
Transcript
00:00:00 Welcome to the Verge cast, the flagship podcast of cancelled out cricket sounds.
00:00:03 I'm your friend David Pierce and I am sitting in my basement about to watch Oppenheimer.
00:00:08 Actually, I should say that differently. I'm about to finish watching Oppenheimer.
00:00:12 I have been watching this movie for months, but something happens every time I sit down
00:00:17 to watch Oppenheimer, which admittedly is like three hours long. So life is busy. It's hard
00:00:21 to find three hours to watch a movie. But every single time I sit down to watch this movie,
00:00:26 somewhere between two and like 11 minutes later, something happens and I have to stop watching it.
00:00:32 It's gotten to the point where Peacock's algorithm took it out of my continue watching
00:00:38 row in the app because I just assumed Peacock was like, there's no way he's ever going to finish
00:00:42 this movie. He can't possibly care. We'll just get rid of it. But I do care and I'm going to
00:00:46 finish it. And today is the day. It's Monday morning, which is a super weird time to watch
00:00:51 this movie, but it's happening. I've got, I think like 40 minutes left. Today's the day.
00:00:56 It's going to be awesome. Anyway, we have an awesome show coming up for you today. We're
00:01:00 going to do two things. First, we're going to talk about active noise cancellation. You might
00:01:04 have seen a video that our team made last summer about kind of the state of the art in those like
00:01:10 over your headphones that try to cancel out subway sounds or airplane sounds or whatever else.
00:01:15 But it turns out the tech and science underneath all of that is actually more advanced than you
00:01:20 might see on some of those headphones. And we have a really fun story about it. Then we're going to
00:01:24 talk about keyboards, specifically iPhone keyboards and the cases that put physical keyboards back on
00:01:30 your iPhone. We're actually doing, I've realized kind of an accidental mini series in the next
00:01:36 several weeks about keyboards, just because I think they're fascinating and we don't talk about
00:01:40 them enough. The ways that we talk to our devices and the tools that we use to do that matters a
00:01:47 lot. Like when, when we switch to voice instead of typing on a keyboard, the way that we use our
00:01:52 devices changes. And when we switch from a physical keyboard to an onscreen keyboard, the way that we
00:01:58 use our phones changes. And so we're going to talk to one person who is trying to put a physical
00:02:03 keyboard back about how that works, what it means and what it changes. All that is coming up in just
00:02:09 a second. But like I said, I have 40 minutes of this movie to watch. No one else's home.
00:02:14 It's Oppenheimer time, baby. This is the Verge cast. We'll see you in a sec.
00:02:17 Welcome back. So last summer, our video team made a video all over New York city in which
00:02:25 they carried around this crazy looking like mannequin skull to test a bunch of headphones.
00:02:30 Noise canceling is hard to explain by just talking about it. So we're going to try to
00:02:34 share with you how it sounds by putting a bunch of noise canceling headphones on the KU100.
00:02:40 It's all about active noise cancellation and whether different kinds of headphones could hold
00:02:45 up. Super fun video. One of the best things we made last year. And I'm told that's actually
00:02:49 only the very beginning of the story of what ANC can do. Will Poore is here to help me explain.
00:02:54 Will, hello. Hello. Will, you made this video last summer. And as far as I can tell, you have
00:03:01 not stopped thinking about it ever since. Is that a fair characterization of what's been going on?
00:03:05 Yeah, kinda. There was an interview we did at the very end of the video that is the thing that
00:03:10 really stuck in my brain. It was with this guy that runs an audio testing lab where we sent all
00:03:15 these headphones and he ranked them for us by their ANC performance. But he also made the point
00:03:21 that the most noise cancellation is maybe not what people actually want. Like maybe we don't
00:03:27 want to walk around in this total vacuum of sound. I think. And he said that in the future, it'll be
00:03:33 about smarter noise cancellation, more user input into what you want to hear and what you don't.
00:03:39 So that's the idea I've been carrying around with me.
00:03:42 Okay, this is a good segue to a question that I have had for a very long time and have been too
00:03:46 afraid to loudly ask in public in front of people. Can't wait. What is active noise cancellation? I
00:03:52 know it's a thing. I sort of know roughly at a basic level, like there are noises and it cancels
00:03:57 them. Yes. What is and there are other kinds of noise cancellation that I also don't understand.
00:04:01 So like, what is ANC? What are we talking about here?
00:04:04 I am so glad you asked. And I brought Andrew along because when we did the video,
00:04:09 I had the same question and I just immediately asked him that.
00:04:12 Okay, good. Andrew Marino, welcome.
00:04:14 Hello.
00:04:15 What is ANC? Teach Will and I, please.
00:04:18 Right. So active noise cancellation takes sound coming from the environment and basically sends
00:04:26 an opposite version of that sound wave to your ear. So when combined, it basically cancels it out.
00:04:33 This is really easy to do with sounds like a loud airplane engine or subway sounds,
00:04:40 noise that's really predictable.
00:04:41 So the two of you have been on this beat for a while now and have been investigating this
00:04:47 question of how far have we come? How far can we go? And where do we actually want to go? Is that
00:04:54 right?
00:04:54 That's right. Yeah. When we started talking and thinking about smart noise cancellation,
00:04:59 I started looking for who's doing this right now. And the first place I looked is Apple
00:05:04 because their AirPods Pro have a bunch of these smart noise control settings.
00:05:09 They call their features adaptive audio. And it's kind of what we're talking about.
00:05:14 It's not like full, let's just cancel out all the sound. It's more, let's make some basic
00:05:21 decisions for you about making really loud sounds more comfortable or dipping your music when you
00:05:27 start talking to someone. I've been playing around with those features a bunch. And in practice,
00:05:32 it's not always super clear what it's actually doing. Like there's no fine tuning. There's no
00:05:38 user input. You just hit the button and it tries to magically like make a better soundscape for
00:05:44 you. So that's the like one place I looked and have been playing with that is out there in the
00:05:50 world. That's a product now that has some smarts, but I also played around with a really early
00:05:57 prototype and I have a bunch of tape that I brought to play.
00:06:00 Awesome.
00:06:01 This is super early stage, a device that could get a whole lot more specific when it comes to
00:06:08 what you want to hear, what you don't want to hear. So I want to play some of this
00:06:13 stuff and we can talk about it.
00:06:14 Yeah, let's go.
00:06:15 Okay, so first listen to this clip.
00:06:18 This is me sitting at a table alone in a room. I'm wearing these very prototype looking headphones.
00:06:30 They've got a lot of extra wires and tape and circuit boards attached to them. There's a vacuum
00:06:35 cleaner running right next to my chair and a few feet away, there's a door and someone outside is
00:06:40 knocking on that door. That sound that we just heard is just coming from a microphone that's
00:06:44 sitting on the table next to me. So that's just what the room sounds like. But while this is
00:06:49 happening, this sound clip, this is what I'm hearing right now. Just the knocks.
00:06:57 That's amazing. This is beyond what we have seen with noise cancelling headphones before.
00:07:03 If you want to just hear knocks, that is not something you can try to get.
00:07:08 You can't tell your AirPods Pro just knocks.
00:07:10 Where did you find this?
00:07:12 So this comes from the University of Washington, specifically their mobile intelligence lab in
00:07:16 the CS department. It's a project that they call semantic hearing. And I got this demo from a PhD
00:07:22 student named Malik Attani. He's one of the people who helped build the system. And he talked me
00:07:27 through how it all works.
00:07:28 There are these noise cancelling headphones, they're Bose QC Ultra. They're going to suppress
00:07:33 all of the sounds. And then what we do is we have these binaural microphones that capture...
00:07:38 Tape to the outside?
00:07:39 Yeah. They capture sound from your environment the way you hear it, because they're on both ears.
00:07:45 And then we pass this through a mobile CPU, an Orange Pi. This thing processes audio in some way
00:07:52 based on the targets that you select, and then feeds it back to the headphones where it's played.
00:07:56 So the starting point for this is a pair of headphones you actually can buy.
00:08:00 Yeah. So the prototype starts with off-the-shelf Bose headphones that use the existing ANC to just
00:08:07 create a clean audio slate.
00:08:09 I see.
00:08:09 And then the system on that Orange Pi listens to the full soundscape separately and picks
00:08:15 out just the specific sounds that it's listening for. In this case, a door knock.
00:08:19 We have a program that runs on the Orange Pi, which continuously just records audio
00:08:25 in chunks of eight milliseconds. Then it passes this to a neural network where it's processed
00:08:32 with the condition that this neural network tries to extract some target class.
00:08:38 This target class is what you select here.
00:08:40 Okay. We've got cat, cricket, dog, knock, and rooster.
00:08:44 Those are the classes that you have queued up right now.
00:08:46 Exactly.
00:08:47 All the most common noises in the world.
00:08:49 Right. Of course. Anything that you would want to be selecting and deselecting.
00:08:52 There's a smartphone screen that I have access to, and there are buttons for each one of those
00:08:57 categories. And in each one of those cases, Malek has fed a neural network a whole bunch of those
00:09:02 clips, different dog sounds and door knock sounds and everything else to train it to recognize that
00:09:07 specific sound. I actually asked Malek to share some of the sounds that he used out of curiosity.
00:09:13 I don't know what I was expecting to learn from that, but here are the cat sounds that he used.
00:09:17 Very familiar with those.
00:09:23 I had to ask because I just pictured him sifting through and he did this. He just sifted through
00:09:28 hundreds and hundreds of sourced cat sounds and cleaned them up and fed them into the system.
00:09:34 So the AI is trained on these categories now so that when I hit the cat button,
00:09:39 the system will mute every sound in the world except for cat meows.
00:09:43 So going back to the demo, I did that. I hit the cat button. Malek turned the vacuum cleaner back on
00:09:49 and he queued up a looping sound of that really annoying cat.
00:09:52 Yeah. Pretend like it's a regular day, lazy Sunday. You want to play with your cat.
00:09:57 But there's a noisy vacuum cleaner in the background.
00:09:59 I hate it when I'm vacuuming and playing with my cat at the same time.
00:10:03 Okay. Transparency mode. My poor cat is trying to get my attention,
00:10:09 but I can't hear him. So I'm going to flip to cat mode.
00:10:13 Wow. That's really impressive.
00:10:20 It's super impressive. The sound is a little odd. To me, it sounds like the cat is meowing in a
00:10:26 submarine, but it's super clear and the vacuum cleaner is totally gone. And for the record,
00:10:31 what you're hearing right now is actually a little bit better than what it sounds like in reality,
00:10:36 because in the room I'm wearing headphones and some of that vacuum cleaner sound is
00:10:40 making it through the headphones. And this is what the headphones
00:10:44 are outputting to your ears. Exactly.
00:10:46 Yeah. Will, you mentioned this is not for actual cats and vacuum cleaners. What is your sense of
00:10:52 what this is actually for? What are the folks working on this thinking they might be able to do
00:10:57 with it? Yeah. I mean, the most obvious example that comes up over and over again is you're
00:11:02 walking down the street and you want to be able to hear a car honking at you because that's a
00:11:09 safety thing. Sure.
00:11:10 But you don't want to hear all the other annoying sounds. Or the folks I talked to brought up
00:11:15 industrial settings where it's really loud and workers are wearing earplugs for safety,
00:11:20 but they want to be able to talk to each other or hear sounds over the intercom or whatever else.
00:11:25 So those are the really obvious use cases. And from playing with this really early demo,
00:11:30 it felt like even this early demo could do that much, which is super impressive.
00:11:35 So you mentioned also like crickets and dogs and other things. What are some other demos that you
00:11:41 were able to test out there? Yeah. So in theory, you can train
00:11:45 the AI on any sound and we played through a few different options that they had queued up.
00:11:52 None of it was perfect. Like there were these times where it was supposed to be
00:11:57 muting all voices, but my voice kind of clipped into the final mix that you can hear. We can play
00:12:03 that. Yeah, that's funky.
00:12:09 Yeah. So, you know, rough around the edges. We went back to the barnyard and listened to all
00:12:14 of those other different sound clips in a lot of different iterations just to see how good it was
00:12:20 at toggling between different sounds. Now that we've done cat and vacuum
00:12:24 cleaner, we'll move on to some other classes. Let's do cat and cricket.
00:12:27 All right. Here's transparency mode. A cat and a cricket are both trying
00:12:35 really hard to get my attention. That's enough out of your cat.
00:12:41 In this demo, you couldn't hear the cat at all. You just heard the crickets.
00:12:45 No, it was gone. Yeah. The big takeaway for me was that this system is already really good at
00:12:51 full isolation. It just needs to make the sounds that you are hearing sound more natural. Like
00:12:57 that's just to my ear. That's where the work was. It was already doing an unbelievable job
00:13:03 at canceling out the stuff you didn't want. I'm desperate to know what the like last demo
00:13:08 he gave you was because I got to assume he's like ramping up in difficulty here,
00:13:12 but there's definitely a point at which he knows this thing will fall apart. Like what was the most
00:13:17 intense demo he gave you? Oh, we did a big finish, which was dog and a rooster and a cat.
00:13:22 This is the most annoying farm in the world. Okay. All right. Just rooster.
00:13:31 It's interesting. Those two, and I could see how subjective this would all be. It feels like those
00:13:43 sounds are bleeding into each other a little bit more. I wonder if they're just a little more
00:13:48 similar than the others. I think it could be that some sounds are similar. Yeah. And finally,
00:13:54 we'll do dog and cat. This is very triggering. This is why I don't like working from home.
00:14:02 Okay. Sorry, cat. Okay. And now I'll just pay attention to the cat.
00:14:10 Okay. This is really impressive, but I'm stuck on two things. One is, is the idea that you will
00:14:21 carry around a smartphone constantly selecting the noises that you want to see? Should I walk around
00:14:27 saying, "Oh, it looks like that cat's trying to get my attention. Let me let the cat in."
00:14:31 Or is the future some sort of beautifully adaptive thing that understands as you're going? Did you
00:14:36 get a sense from them how they see this eventually working? Yes. And I talked to the head of this lab
00:14:42 all about that. But before we did that, we talked about how they're going to get to that point.
00:14:47 That was my other question is how do you take this from a pair of Bose headphones
00:14:51 with a bunch of junk taped on it to an actual thing that humans can buy?
00:14:56 So yeah, let's talk about their full roadmap from where they are now to what you're describing. The
00:15:01 first step in that process is just getting better at doing all of these tasks. And that means
00:15:08 feeding their model a whole lot more data. More cricket sounds.
00:15:11 So many more cricket sounds and grasshopper sounds and cicada sounds and just basically
00:15:16 train it to recognize a million more categories of sounds and to recognize all those individual
00:15:21 sounds just a lot more accurately. And if they can do that and do that super well,
00:15:26 then the possible use cases just explode. You could totally see this being applied to
00:15:33 different tasks. Like say, I want to not just listen to sounds, but specific people.
00:15:39 Like I want to listen to you in a conversation so I can maybe have some representation of how
00:15:44 you sound like and then I can just pick up your sound as we speak, as we have a conversation,
00:15:50 even if there's lots of other people around us. So all the current headphones right now,
00:15:55 they do it really well because they can predict when these sounds are coming.
00:16:00 So when you're in the airplane, you know these sounds are going to be 10 seconds from now
00:16:04 in 10 seconds before. But a bird sound, a cat meowing, crickets chirping, you don't know when
00:16:10 those are coming. So how would they be able to predict? They would have to move pretty fast
00:16:15 on the device. Exactly. That's the problem of what they call transient sounds. And that's a
00:16:22 big hurdle for existing ANC technology. This system is really good at it, but it needs to
00:16:29 process the incoming audio super, super fast. There's no time to send it to the cloud for
00:16:35 processing. There isn't really even time to send the incoming audio to your phone's processor over
00:16:41 Bluetooth. You would just end up with a lag between what you're hearing and what your eyes
00:16:47 are actually seeing, which is just a complete non-starter. So all of the work that we're
00:16:51 talking about doing has to happen on board the headphones themselves. If we can get this model
00:16:57 to be much, much more efficient through lots of just standard machine learning techniques,
00:17:02 then we can fit a more capable model into this small device. Another way to also help on the
00:17:08 hardware side is in the recent few years, there's been lots and lots of companies that work on
00:17:16 designing AI chips that can do a large, large amount of computation with very low power
00:17:22 consumption. And those things are very promising. So you've just made me imagine what amounts to
00:17:28 like a pair of headphones with a GPU attached, which I think is like literally what he's
00:17:34 describing. And that both seems cool. And like, maybe where all this is going, but also very
00:17:39 expensive, very big, and maybe very far away. What is the sense of how we get from literally
00:17:47 strapping like an NVIDIA GPU to my face to the thing he's describing? Which I think ultimately
00:17:53 it has to look like a pair of Bose headphones, right? Like we're not going to take anything
00:17:56 that doesn't look like a pair of Bose headphones. This is where we'll have to see. They're talking
00:18:00 about a hearing aid sized form factor. They want to go much, much smaller. And they think that
00:18:08 they can do it with these purpose built lightweight chips that are really good at running AI systems.
00:18:15 So they're playing with all of the chips that are very new and newly available. And they're sort of
00:18:20 doing all of these different things at the same time. They're doing the software optimization,
00:18:25 they're feeding the model all this more data, and they're working on the hardware too,
00:18:30 to shrink it as quickly as possible. And they think they have the raw materials to do that
00:18:34 pretty soon. So we have like an academic lab masquerading as like a full stack electronics
00:18:39 company here. Yeah, absolutely. That's the fun thing about this lab in particular is there.
00:18:43 That's very holistic. Oh, what about the user interface? Am I going to be like asking Siri
00:18:49 every 10 seconds? Like, did you hear my cat? Is my cat meowing? So they're thinking a lot
00:18:55 about this. Because even when I was playing with this demo, I was like, Oh, God, five buttons,
00:19:01 this is just five buttons. In the real world, you would need 5 million buttons to actually dial in
00:19:06 the soundscape that you want. So obviously, you need some sort of automation here. And the lab
00:19:12 is looking at different AI assisted ways to do this. So this whole thing is going to be AI's on
00:19:17 AI's. But I talked to the head of the mobile intelligence lab, Shyam Golakota, about exactly
00:19:23 this. We're going to have some systems in the next few months to a year on trying to go away
00:19:29 slowly from like actually having the user to like pick different kinds of sounds to less and less
00:19:35 involvement of the user itself. You can imagine a system where the AI automatically learns what
00:19:40 you want to listen to. And I think that's something which we are building towards. And I
00:19:43 think by the end of the year, hopefully, we can actually have a system where the AI would
00:19:47 automatically learn what you want to listen to. That is ambitious. Yeah, yeah, they're juggling a
00:19:53 lot of balls at the same time here. And the you know, the end goal that Shyam is envisioning is a
00:19:58 system that is using data from, say, your smartphone to think and learn about where you are,
00:20:05 what you're doing, and start to make those decisions kind of with as much or as little
00:20:11 input as you want as to what you're hearing and what you're not. So say, you know, again,
00:20:16 you're living in a busy city with lots of loud traffic, but you're inside your house.
00:20:21 So that's exactly the thing. Like, clearly, an AI can tell depending on my motion and my
00:20:26 IMU, whether I'm moving as much, where am I, it knows my GPS location. So it knows that I'm in a
00:20:32 house. Does it really need me to listen to like the sounds of the honks? Not really. But if I'm
00:20:38 on a street, maybe it can because I'm walking on the street or I'm in a car, it should potentially
00:20:44 be like, you need to hear these honks. Although then the next place my brain goes is how to make
00:20:50 sure that it's doing that safely and reliably. Because there's one thing we've watched from
00:20:54 Tesla autopilot to, you know, chat GPT hallucinating to whatever else. It's like,
00:21:00 there's an asterisk on so many of these tasks that we just sort of let the AI take care of.
00:21:07 And it strikes me that trusting your headphones to decide when you need to hear a car horn or not
00:21:13 could be a high stakes situation. Yeah, yeah, yeah. I think that's definitely something which
00:21:19 has to be integrated in these systems. And it could be that certain things like if I'm actually
00:21:23 in the house, that's when you remove all these ambient sounds. But it's just like hard coded
00:21:28 saying that you cannot remove certain things in outdoor environments where you know that it's
00:21:32 important. But that's an extremely important point, which is like when we deploy these things
00:21:35 at scale and have this thing automatically learn, you got to have certain rules which
00:21:40 prioritize safety over functionality, I think. The thing that's being described here is so
00:21:46 beautiful and smart and so non existent in the real world that it's like it's like what we talk
00:21:51 about with self driving, right? Like if people weren't weird and didn't make strange decisions,
00:21:56 and pedestrians and bicycles didn't exist, self driving cars would be awesome. So easy, so great.
00:22:01 But the world is a weird, messy place full of edge cases. Yeah. And that's where we just haven't
00:22:07 seen that whole half of this prototype. They have proven out that they can make crickets go away,
00:22:13 which is impressive, which is impressive. And it's like you can see a roadmap to extracting
00:22:17 a million other sounds based on that. And then this bigger end state of dialing in exactly what
00:22:24 you want is the loftier thing. And there's there's a lot of work to be done there. But if they can
00:22:30 figure out even some of this stuff, you can imagine, you know, them getting to a really
00:22:36 interesting product. Xiom is interested in smart hearing aids. You can see how anyone with a
00:22:43 hearing aid could use even a portion of this technology. Anyone with a sound sensitivity.
00:22:50 I use noise cancellation just to turn the world down to keep my anxiety in check. So I would love
00:22:56 to be able to dial in that sound a little bit more. Xiom made the argument, too, that a lot of
00:23:03 us just live in a very noisy world and noise just as a thing that we're all confronted with is bad
00:23:09 for everybody. And it actually does impact your health. There are studies which say that noise
00:23:14 pollution affects your sleep and that ends up affecting your lifespan. So I do think that
00:23:21 having this kind of a capability where people can control their environment is just not a toy.
00:23:26 I think it actually can have a meaningful impact on the quality of life and sleep. And a lot of
00:23:32 things which I think are extremely important for humans to thrive and not just survive.
00:23:37 So, I mean, wherever this specific prototype lands, we're definitely leaving the fancy earplugs era
00:23:44 of noise cancellation and entering an era that is much more specific and bespoke and a little bit
00:23:51 smarter. What that looks like TBD. But I am very excited about that concept.
00:23:56 I am, too. And I wonder, Andrew, just to come all the way back around to this idea of what we were
00:24:02 talking about at the very beginning with what active noise cancellation does. It seems like
00:24:05 knowing this kind of stuff where you can say, like, not only is this an external sound and I
00:24:10 want to get rid of it, but I know specifically which external sound it is. I think even leaving
00:24:15 aside sort of that big, giant, grand vision, it just feels like that step alone does a lot.
00:24:21 Oh, totally. I mean, to be able to walk into a store and the headphones know I'm in a store
00:24:28 and I'm going to be talking to someone checking out, or I have a library of sounds that are unique
00:24:35 to my experience that I would like to tune out and the sounds are only at this specific coffee shop
00:24:41 that I go to study. That's super compelling to me. Do I want my headphones to know where I am at all
00:24:48 times? I'm not sure. But there's a lot of awesome possibilities of just tuning things out and
00:24:55 tuning things in wherever you are. Yeah. I think the self-driving example just keeps feeling more
00:25:02 and more true to me as we talk about this, that the first thing where it's like, okay, my car can
00:25:07 park itself and will stay in its lane if I try to drift out of it because I'm not paying attention,
00:25:12 right? Doable. Awesome. If you extend that all the way out to all the cars on the road will drive
00:25:17 themselves and robot taxis will be everywhere. Like you've sort of lost me. Maybe it'll happen
00:25:21 someday in the future, but that doesn't make the first thing any less cool. And I feel like there's
00:25:26 a lot of that in here too, where it's like, even just boil it all the way down, this is going to
00:25:30 make noise cancellation better because the cats get through my headphones now and they can stop
00:25:36 getting through my headphones once we get this tech. Totally. And I think that no matter how far
00:25:41 into that idealized end state they get, they're very confident that they can get a lot farther
00:25:47 very quickly. They have access to these chips. The software is improving very quickly. So they
00:25:54 have a lot of ambitions for this year. They think they can get to a much, much smaller form factor
00:25:58 by the end of this year, for example, and they have a lot more very specific ideas for that sort
00:26:05 of filtration step. So yeah, it's, you know, for me, that very, very limited demo was impressive
00:26:12 enough and they had enough plans for this year that I'm going to be following that lab and see
00:26:19 what they come up with next. Because I think it's going to be an interesting year for it.
00:26:22 >> Do you think at the end of this, they're going to try to sell me headphones? Like,
00:26:25 what's their game plan here? Are they going to spin out a company and try to be a headphone
00:26:29 player? Are they going to try and work with some of the folks out there? Did you get any sense of
00:26:33 how they want to get this out in the world? >> They seem to open to that. You know,
00:26:36 they're an academic department. They're talking about commercialization. You know,
00:26:40 they're like a lot of academics. They're walking this fine line between, you know,
00:26:45 just moving this technology forward and experimenting with it, but also like
00:26:49 knowing that they've got something and they feel like they are potentially ahead of a lot of big
00:26:54 tech companies on this. So they are thinking in those terms. And I think that's one of the
00:27:01 reasons they're trying to work as quickly as they are. >> Yeah, it does seem like this is happening
00:27:04 fast. I feel like, Andrew, you were the one running around New York with the head on the subway a year
00:27:10 ago. This strikes me as something that a lot of those companies building ANC headphones are
00:27:16 probably also thinking about. Like, do you think this is the next turn for a lot of those companies?
00:27:22 And is this going to be the next thing that makes these headphones better for people?
00:27:25 >> I think so. And I know Bose is thinking about it a lot as the next generation of
00:27:30 noise cancellation. Apple is already trying to do the adaptive things, as we mentioned.
00:27:35 But right now, it's just boosting things instead of taking things out. So if there's a world where
00:27:41 they can remove the dogs out of Will's room while he's recording a podcast, I mean, there's like
00:27:48 endless possibilities. >> There was a dog button on that prototype. We just need that prototype,
00:27:52 and we'll be fine. >> Well, by this time next year,
00:27:55 the Verge cast will be a listenable podcast. It's going to be great.
00:27:57 >> Thank you and sorry. >> All right. We got to take a quick break,
00:28:03 and then we're going to talk keyboards. We'll be right back.
00:28:05 All right, we're back. If you remember the smartphone world before the iPhone,
00:28:16 it was much wackier than it is today. Back then, there were lots of ideas about how a smartphone
00:28:22 should look and work. Some of them twisted open, some of them slid open, some of them flipped open,
00:28:27 and a lot of them had physical keyboards. And then the iPhone all at once kind of upended that.
00:28:33 Here's Steve Jobs at the original iPhone launch in 2007,
00:28:37 explaining what he saw as the problem with hardware keyboards.
00:28:40 >> Here's four smartphones, right? Motorola Q, the BlackBerry, Palm Treo, Nokia E62,
00:28:46 the usual suspects. What's wrong with their user interfaces? Well, the problem with them
00:28:51 is really sort of in the bottom 40 there. It's this stuff right here. They all have these keyboards
00:28:57 that are there whether you need them or not to be there. And they all have these control buttons
00:29:03 that are fixed in plastic and are the same for every application. Well, every application wants
00:29:08 a slightly different user interface, a slightly optimized set of buttons just for it. And what
00:29:14 happens if you think of a great idea six months from now? You can't run around and add a button
00:29:18 to these things. They're already shipped. >> It's a good take, right? And history
00:29:22 proved him pretty much correct, which is why virtually every phone on the planet now has a
00:29:27 virtual on-screen keyboard instead of a big physical one underneath. Jobs did definitely
00:29:33 have it right, but not all right, because there are things I think a lot of people miss about
00:29:38 physical keyboards. You can type faster on physical keyboards. That's just true. There's
00:29:43 plenty of evidence for that. You can type without looking at your phone so much. You can type
00:29:47 without the keyboard taking up like 40% of the screen. And so over the years, various people
00:29:52 have tried to bring back the keyboard to the smartphone. Remember way back when Ryan Seacrest
00:29:58 was out there hawking a case called Typo that put a keyboard back underneath your iPhone?
00:30:02 This is how he explained it in an interview with TechCrunch.
00:30:04 >> This has changed my life. I literally was suffocating thinking about how I was going to
00:30:11 work every day because this is my office. The tablet and the phone, that's how I work.
00:30:17 I have incoming emails from everybody that I work with and for my partners, and this is how I
00:30:21 respond and can respond quickly and be able to actually type more than two sentences.
00:30:25 >> I get it, Ryan. I feel that too. But Typo didn't really work out, and none of the other
00:30:31 projects trying to do this did either. And yet now, all these years later, there's another company
00:30:36 trying to pull it off. It's called Klix, and it's co-founded by a guy you might have heard of.
00:30:40 >> I'm Michael Fisher, aka Mr. Mobile or Captain Two Phones. Take a break.
00:30:45 >> Michael's a longtime creator and phone reviewer, and he knows this space better than most,
00:30:50 which made it frankly all the more surprising that this is a thing he decided to do.
00:30:54 So I brought him onto the show to figure out whether this idea has a chance to really succeed,
00:30:59 why he's convinced there's still room for a physical keyboard on your iPhone,
00:31:02 and what it takes to make a really great tiny little keyboard. And we just started the story
00:31:07 at the beginning. Why would you, in the year 2024, set out to build a physical smartphone keyboard?
00:31:13 >> Oddly, the thing that was the germ of Klix came to me. It didn't originate from my brain.
00:31:20 I got a call from a guy you probably are familiar with, Crackberry Kevin.
00:31:24 It's a guy who helped me build Mr. Mobile in the first place, kind of like one of my closer
00:31:28 friends in the business. And he said, "I got to show you something. I will absolutely kill you if
00:31:31 you show anybody else what this is." And he sends me this 3D render. And you know, you and I are in
00:31:37 the same business for a lot of my day. You know what it's like to get something in your inbox
00:31:41 that's like, you look at it, you take a beat, and you say, "Oh my God, you crazy bastards,
00:31:46 you actually did it." So I think that's what I said out loud. But it was just a render. It wasn't
00:31:51 even a product. It was just this very obviously an iPhone inserted into a case with a QWERTY
00:31:56 keyboard that looks nothing like it currently does. And I said, in a profound demonstration
00:32:01 of how little business foresight I have, I said, "You know Mr. Mobile has to have the exclusive on
00:32:05 this, right?" >> That's good. Always think in one step ahead, I guess. >> What an idiot. So
00:32:11 within a week though, you know, I'm giving so much product feedback. We're talking all the time.
00:32:14 It's very obvious that we should be working together on this. Turns out that's what Kevin
00:32:18 had in mind the whole time. So I come aboard, we finish building the company with the guys who,
00:32:23 full credit, deserve credit for the initial germ of this idea in the first place, the veterans
00:32:27 from FX Tech, the last company to make a truly great Android sliding keyboard smartphone. And
00:32:33 that was, dude, that was April of 2023. >> Oh wow, that was fast. >> Yeah, by January at CES,
00:32:38 we had a product that we were able to, we were ready to ship. So it went fast is right. It was
00:32:44 insane. >> What I like about that though is that is a group of people who are maybe most prone to
00:32:49 being incorrectly nostalgic about keyboards, right? You have some of the world's foremost
00:32:55 BlackBerry people. You have you, who has just never stopped making videos about the Palm Pre.
00:33:00 >> True. >> And you have a bunch of people who have made these things in the past. So how do you
00:33:07 check yourself against, like, boy, I wish the world still existed in which we wanted keyboards,
00:33:12 versus like, this is a thing we should do and bring it back? >> So... >> Or maybe you don't,
00:33:16 you just soldier through and hope other people feel the same way you do. >> I think there was
00:33:20 an element of that, really. I mean, one of the things we knew is we had a lot of smart people
00:33:23 on board, right? And I won't give you the whole marketing pitch of everything, but we really do
00:33:26 have the best people in the roles that need to be within the company and the most expertise there
00:33:30 is. And at no point, you know, did we check ourselves all the time? Yes, but at no point
00:33:34 did anyone say, you know, I don't think this is the time. Really, what I had to be sold on this,
00:33:38 because I was a little, I turned on my skepticism node for a hot second. I said, guys,
00:33:43 is this it? Like, no young person is going to even want a single physical button on this thing.
00:33:49 They're like, what's going on? And it turns out that the time was right, not only for a variety
00:33:54 of unrelated reasons, but because you can introduce this as a new concept to somebody like,
00:33:59 not to name drop, Max Weinbach, who was like our most difficult briefing at CES. He was like,
00:34:03 he's like 2022. And he's just like, I think this is ridiculous. But then you show him what happens
00:34:08 when you are using, you know, Instagram live and you have your entire screen there and you go to
00:34:13 type something and the virtual keyboard doesn't eat up 40% of it. Or you show somebody a keyboard
00:34:18 shortcut and you're like, wait, what did you do? I'm like, well, I just hit the spacebar and it
00:34:21 advanced the screen for me or I hit, you know, whatever, drop the notification shade. That's
00:34:25 part of the fun. All the BlackBerry people know this stuff already. All the, you know, the people
00:34:29 who had a Palm trio, like you and I maybe, uh, know all these things and to introduce it to
00:34:35 somebody who was not alive when the, uh, when the BlackBerry was a thing, it's pretty cool.
00:34:40 That's interesting. So we're actually like past nostalgia for a lot of people where it's so far
00:34:45 gone that it's actually a new thing again. Yeah. It's a full circle. Exactly right.
00:34:48 But I would imagine for you as somebody who has used these things, but not built them before,
00:34:52 part of the process is just sort of learning how to learn a keyboard and like the vocabulary of it.
00:34:58 And what it means for a keyboard to be good or bad. Cause I think as a user, you have this
00:35:03 instinctive feel, but I have to imagine that at some level there is like an objective measure of
00:35:08 a good keyboard that you have to learn how to reckon with in this process.
00:35:12 There is. So like the first prototype of clicks I got to use had these domed transparent key caps.
00:35:19 So we printed the characters below the surface of the key, like a Palm pre, right? And because I'm
00:35:25 so in forever in love with the Palm pre until I die, I loved it. And I was like, Oh, this looks
00:35:30 so good. And this is the language we should be going for. And perfect. Well, it turns out though,
00:35:35 that when you make a domed transparent key, that is perfectly circular in the design we were using,
00:35:40 it looks great from head on. And then you tilt that phone by 20 degrees and you can't tell what
00:35:46 the letter is because there's visual warping, right? There's this skewing that happens. You're
00:35:50 like, is that a semi-colon or a comma? I don't know what I'm doing anymore. So very quickly,
00:35:54 we had to make little decisions like that. Like, well, let's see what flat polycarbonate keys with,
00:36:00 you know, satin printed top layer paint looks like. Well, it turns out that's a lot better.
00:36:04 So there's kind of major changes early on. What I found surprising is you and I,
00:36:09 we've covered phones a long time. You know, how early on designs get locked in and you see people
00:36:15 talking about why didn't they just make this change? Because we were giving them feedback
00:36:18 after the, after the unveiling. And it's like, well, because they were been sitting at retail
00:36:22 boxes for three months at this point. With something like Klix though, we were making
00:36:26 changes right up until the end. At CES, we got the final keyboard while we were at the show.
00:36:31 The thing we showed everybody was kind of like a second to last revision. The final keyboard came
00:36:36 in. It wasn't even built into the case. We're carrying around like a, like a component with
00:36:40 no paint on it or anything. It's just like, okay, don't look at this. This is gross, but just close
00:36:43 your eyes and feel it. And everyone we showed that to, we got the final sign off on it. Kevin
00:36:48 was so excited. We were like, oh, that's it. Every nerd we trust has signed off on that. So dude,
00:36:53 I'm going forever. We had like a hundred revisions. It was crazy.
00:36:56 - I will say I am most fascinated by like the last handful, because I think I would assume,
00:37:01 and please correct me if I'm wrong, but that you get through kind of 80% of the problem
00:37:06 fairly quickly. And you, you settle on like the rough idea of the thing that you want.
00:37:10 And then you spend like the rest of your life until the heat death of the universe debating
00:37:15 like tiny millimeters of everything. But that stuff does matter in these really like hard to
00:37:20 understand ways. So like the last 10 revisions, I think must've been so interesting and weird
00:37:25 to go through. - They were so wild because you're fighting, you're trying to integrate your,
00:37:29 your personal history without being a, you know, without being slavish to it. Right? So a good
00:37:34 example is the shift and alt keys right up until close to the end, they were swapped. And I kept
00:37:39 trying to switch to alternate characters and I kept messing it up. And I'm like,
00:37:42 all right, is this muscle memory or what? And we talked about it. We kind of went back and forth,
00:37:46 got the whole team involved. And it was like, no, we got to swap those keys. And once we did
00:37:50 everything felt right again, because what do you know? You find the corner more easily.
00:37:53 My least favorite change, I think I talked about in my video was I want that old number pad. I want
00:38:00 the, you know, the, the three by three BlackBerry Palm Trio kind of number pad on the keyboard.
00:38:05 Somebody much smarter than me on the team said, look, if we're designing the best iPhone physical
00:38:09 keyboard, you got to design it like the iOS virtual keyboard. You got to give the number
00:38:13 row up top. And I said, I hate that. You're right. I hate it, but you're right. I can't say you're
00:38:18 wrong. That's what we got to do. So little tiny changes like that. Have you gone back and retested
00:38:24 all the old physical keyboards? It's so weird because I was using the Palm pre obviously,
00:38:29 most recently. And I was like, I remember writing for this keyboard back in the day when I owned it
00:38:34 in 2009. And I pick it up now and I'm like, Ooh, those keys barely move at all. And the Centro,
00:38:40 right. Where I was like, I could type fine on this in 2008. Nope. Even in 2008, I was like,
00:38:44 this is too small. I couldn't do it. But you could use the stylus on the buttons though,
00:38:49 which was great. That is true. But other surprises in the other way, I'll get off webOS in a second,
00:38:53 but real quick, the HP Vier, world's tiniest keyboard phone that a lot of people made fun of
00:38:59 very good keyboard for the footprint. And for me, best keyboard I ever used, but did not own
00:39:05 Sidekick 2. Oh yeah. They're like lovely clicky travel and feedback. And also covered in that
00:39:11 delightful silicone rubber that held onto your thumbs. And oh man. What's your favorite keyboard
00:39:18 of all time? If you don't mind me asking. I think if you made me pick two, it would be the Trio 650,
00:39:23 which I loved to pieces. I loved that phone and the Motorola Q, which I actually think if you made
00:39:29 me pick one is probably my answer. That keyboard was amazing. I love it. You know why I think the
00:39:34 Q worked is because you had this nice interplay of materials. They were plastic hard keys,
00:39:38 which doesn't sound nice, but the phone was soft touch. That's one of the things we were,
00:39:42 we were doing on clicks actually pretty late was once we had the keys locked in, I was like,
00:39:46 something's still weird and missing about this. And it turned out it's what's around back.
00:39:50 So we ended up putting that stamp, like vegan leather back plate on there to make it feel,
00:39:55 you know, to give you the entire tactile experience. Cause you got, you know,
00:39:58 eight other fingers that you got to worry about when you're typing. Right. What do they feel?
00:40:02 But if you look at those keys in cross-section, you'll notice they're domed over the top too.
00:40:06 And that helps with, with determining separation of keys by touch. So you'll always have an easier
00:40:11 time on a domed keyboard than you will on something where it's all crowded together.
00:40:15 Like on the bold and the bold didn't do that because it was a better typing experience.
00:40:18 The bold did that because BlackBerry wanted a narrower phone. So it was, that's why you had
00:40:22 those triangular frets in the corner. So they're like the science of this stuff is fascinating.
00:40:27 And I'm so glad it's not lost to history because you can still talk to the people who designed it
00:40:31 and be like, Oh, this is why you did this. These are smart ideas. It's good to bring some of them
00:40:35 back. Tell me also about doing this in kind of the iPhone era, because I think people have tried
00:40:40 and failed at this in part because the iPhone it's good at what it does already. Yes. How do you like
00:40:45 put those two things together? First of all, one, frankly, made our jobs a lot easier is that we,
00:40:52 you don't have to do anything with clicks. You don't have to install our app. If you don't want
00:40:56 to, you literally plug an iPhone in and the iPhone goes, you know, which uses a lightning
00:41:00 port or the USB port. And the iPhone says, Oh, that's a physical keyboard because all the hooks
00:41:04 that it needs are built into iOS to begin with. So you just start typing and it's like, okay,
00:41:08 well, it turns out this product can handle it very well. And it's not nothing that that's the case,
00:41:13 by the way. And I think, I think a lot of people would be surprised that that's the case because
00:41:17 there just isn't really a reason for the iPhone to do that, but it is surprisingly receptive to
00:41:22 the existence of a physical keyboard. I think in terms of like a philosophical thing, I had
00:41:28 a little bit of the same problem grasping it at first. It was like, gosh, what are we building?
00:41:32 And then somebody said something very smart. And they said, look, nobody looks twice. Nobody
00:41:37 thinks twice when you use an iPad with a magic keyboard. And then you say, well, I don't really
00:41:42 need the keyboard right now. So you take it off. And then later you're going to write an email and
00:41:45 you put the keyboard back on. Like that is the definition of normal use case for that product.
00:41:50 Why shouldn't the same thing be true of the phone? And the minute somebody put it in that lens for
00:41:55 me, I was like, oh, okay, good. And, you know, we kind of hedge a little bit too, because there's a
00:42:00 button on clicks you can hit to call up the virtual keyboard. So like, you know, there are times, man,
00:42:05 where I'm walking around and I'm like, I do not want to type on a physical keyboard with one hand.
00:42:09 It is annoying and slow and I don't like it. So that's the virtual keyboard. And I can swipe
00:42:12 around and stuff and I can use emojis and then I can put it away. So like, it's really about
00:42:17 the adaptability for me. Like, you know, it's eliminating buttons. Yeah, good call. Had to do
00:42:22 it. But turns out there's a lot that you miss. So why not just have buttons when you want them?
00:42:26 That was kind of our guiding philosophy after we figured out that we could actually make it.
00:42:30 What have you seen in terms of like, surprising, cool stuff that you can do on the iPhone? I mean,
00:42:36 one of the things, you know, you've mentioned and I've seen you talk about elsewhere is just
00:42:39 added screen real estate, right? Like, you don't realize how big your phone screen is
00:42:42 until you can actually see the whole thing. It's so annoying. Which I think is real. I also wonder,
00:42:48 like, you've been using these things longer than just about anybody, right? Like, what have you
00:42:51 seen in your own kind of phone usage life that is helped by a physical keyboard like this?
00:42:55 Yeah, one of the, so the example that they make me do, that my colleagues make me do on video,
00:43:00 because you're the creator, you're the influencer, do the live Instagram streaming thing and typing
00:43:05 while you're doing, I'm like, I don't, I don't care about that. What I care about is I'm in 85
00:43:08 group chats, right? And when you're in a group chat, and you're looking at all the context of
00:43:12 what people have said before, and then you go to respond, and that virtual keyboard pops up,
00:43:16 and then it clears half the screen, stop. So I love using clicks for that specifically. But it's
00:43:22 kind of like an Easter egg hunt. There are all the hotkey shortcuts that work on the magic keyboard,
00:43:26 or at least a lot of them work in clicks too. So there's that. And I go into every app, and I try
00:43:31 the old, the old BlackBerry shortcuts. Can I hit T to go to the top? No, I can't. What does spacebar
00:43:36 do? Oh, that scrolls. Okay. Well, this phone is really tall. And now with clicks, it's even
00:43:40 taller. So how do I get to the notification shade? Does globe N work? Weirdly, yes, it does. And
00:43:46 what about shift space? Oh, that goes the other way. And like, just there are so many little old
00:43:51 habits that I sort of forgot about from the days of reviewing all the old QWERTY phones. And again,
00:43:56 to iOS is great credit. They didn't rip any of that stuff out. It's all just sitting there,
00:44:01 waiting to be utilized. My favorite is globe N. You know, Safari's got a ton of them,
00:44:05 but I mostly just use the spacebar to scroll because I'm lazy.
00:44:08 Does it have the thing that I love so much about the BlackBerrys, which is that if you're on the
00:44:12 home screen, and you just start typing, something happens? So on iOS, you have to do, you have to
00:44:17 invoke spotlight like you would on a MacBook. So you do command space, and then you pop up. So
00:44:21 Craig Federighi, if you're listening, and I know that you are, just get rid of that. Just let me
00:44:26 start typing, and it should pop up the spotlight. Right. Do you remember that on the BlackBerry?
00:44:31 It was the best. It was the best. You're like, "I want to go to my calendar. I'm just going to
00:44:34 pull my phone out of my pocket and hit C-A-L, and it's going to take me to my calendar." But I think,
00:44:37 I mean, that comes back to kind of the question about, you know, all the other affordances we've
00:44:42 made for the ways that people use phones now, right? Like, I think the swipe typing is an
00:44:45 interesting one. And my sense is that has kind of come and gone. And just purely anecdotally,
00:44:50 everybody I know who uses Android knows about the swipe typing and uses it, especially when you're
00:44:55 using your phone with one hand. No one I know who uses an iPhone even knows that that's a thing you
00:44:59 can do. And like, the spacebar trick to move the cursor is my favorite party trick, because no one
00:45:05 knows that's a thing. And you show them, and it just blows their mind to pieces. But that's the
00:45:09 kind of stuff that all of these companies have spent this much time working on because they
00:45:13 didn't have physical keyboards. And in a funny way, they're trying to solve for the problems
00:45:17 of physical keyboards. But also, like, once you got rid of them, you had to figure out a better
00:45:22 way to move a cursor because you didn't have arrow buttons anymore. And like, you know,
00:45:24 what's great is arrow buttons. But now we've sort of come past that. And so now you have to go back
00:45:29 and say, okay, we have a physical keyboard. How do we make it do some touchscreen things?
00:45:34 Yes.
00:45:34 Which is an interesting challenge.
00:45:36 If you're utterly insane, how do we make an individual tiny touchscreen behind every button?
00:45:41 Which is what I want to do. I want you remember the Samsung alias?
00:45:45 I do. Yeah.
00:45:45 Yes. One of those. Let's put an E Ink screen behind it. Let's put an OLED behind every one
00:45:50 of these. Let's make a $500 clicks and see if anybody buys it. It'll be great.
00:45:54 That's the full lunacy keyboard thing that I'm very excited for.
00:45:59 I'm going to make that and then I'm going to retire and move to Mexico.
00:46:02 Why start on the iPhone, by the way, as opposed to starting with, you know, Galaxy phones or
00:46:06 Pixel phones or trying to do something that is sort of universally Android? Like, I've been
00:46:10 spending some time with the Backbone folks, and they've been doing some really interesting work
00:46:13 to try to figure out how do we make one thing that works for everybody. And I can imagine,
00:46:17 you know, if you build something that works for everybody, huge increase in the total
00:46:21 available market. But also, if you build something for Android, you are more likely to hit the sort
00:46:26 of phone nerd people who care about this stuff the way that you and I do. Why not start there?
00:46:31 It's really about numbers. You know, honestly, the total addressable market of iPhone users
00:46:36 is something like 1.3 billion. And we know that clicks is not for everybody. It's always fun when
00:46:42 somebody offers up some wisdom to me on social media, like, "Well, you know, this is... No one's
00:46:47 going to buy this." I'm like, "Okay, if one in 1,000 people buy it, and you're targeting a base
00:46:52 of, like, of iPhone users, like, just from a sheer numbers perspective, it's... I have preferred
00:46:56 Android for a long time. Partially because of the hardware diversity of that ecosystem, right?
00:47:01 So I want to build one eventually. But again, you got to work in stages, as you well know. You have
00:47:06 to start, build a great product, build a healthy business on the back of that, and then you can do
00:47:11 other things. To your other question about, like, why not make a one-size-fits-all one,
00:47:15 definitely considered that. I mean, we got close to announcing that. It'll probably come in the
00:47:20 future. There were just too many compromises. It doesn't look as good. You can't get as many
00:47:26 cool certifications when it's not built for a very specific type of iPhone, right?
00:47:30 Sure.
00:47:31 And I hate to keep harping on the aesthetics, but we really wanted to ship something that wasn't
00:47:35 this kind of kludgy... I won't name-drop any previous efforts, but, like, you know, something
00:47:39 that just doesn't look very good, doesn't feel like it was purpose-built for the phone you spent,
00:47:44 what, $1,100 on. So let's say, you know, launch goes well, everything comes out over the next
00:47:50 few months. What's next for Kliqs? Like, I know I've seen you're going to do... your plan is to
00:47:54 do more colors and do drops and that kind of stuff. But I also feel like the thing that I've
00:47:58 learned about hardware startups over the years is that getting the first one out the door is
00:48:02 borderline impossible. It's a huge amount of work. You have to talk to a bunch of manufacturers and
00:48:06 companies that don't know you and don't care about you. But then if you can do the thing,
00:48:12 prove you can do it, prove you're serious, like, the world kind of opens up to you a little bit.
00:48:16 And you strike me as somebody who has some pretty wild ideas about where all of this could go.
00:48:21 Do you feel like there is tons of runway left in, like, the footprint of a keyboard? Do you
00:48:26 have big ideas about other stuff you want to add in? Like, tease the future a little bit for me.
00:48:30 This is not true what I'm about to say.
00:48:33 Perfect.
00:48:37 But to an extent, you know, I almost wish the beginning portion was as harrowing as you
00:48:44 described. But because we have so many people in the right places, we don't have as many of
00:48:49 those problems as we might otherwise. So we've had to think ahead of time, ahead of when I would
00:48:54 prefer to think about these larger conceptual questions, these vision questions of, like,
00:48:59 at some point you have to ask yourself, well, what are we going to do? Like you just said,
00:49:02 are we just going to own buttons on mobile phones? Or are we going to be the kind of company that
00:49:07 asks, well, what else can click? Look, I have giant, crazy, dumb ideas in the latter category,
00:49:15 and we have a very solid opportunity to make a pretty good business in the former category. But
00:49:20 I'll be honest with you, it's not clear which direction we will go, because we still do have
00:49:25 to focus on getting, you know, we've shipped Founders Edition. That's thousands of units,
00:49:29 but it's not the main production run. So we're going to take the time that we have right now
00:49:34 to deliver the best possible Gen 1 product to all the people who want to buy them, and then
00:49:39 we'll have a better answer for you this time next year.
00:49:41 That's a very good and very diplomatic answer.
00:49:43 Thank you.
00:49:44 And I appreciate that. That is someone who has been in a company for a while,
00:49:48 and I'm proud of you for that.
00:49:49 Thanks, man.
00:49:50 Okay, so I won't ask you to tell me the future. Give me one huge, wild idea that there's absolutely
00:49:55 no chance you're ever going to build, because it's impossible for one reason or another.
00:49:59 You remember, there was an old Kyocera that ran Palm OS, didn't even have a color screen,
00:50:02 we're talking 2001, I think it was a 71 something, and it had a touch screen,
00:50:08 but it had a stowaway detachable numeric keypad that attached via a hinge on the bottom,
00:50:13 and if you didn't want to take it all the way off, you could flip it around back so it would
00:50:16 get out of the way.
00:50:17 This is a very 2001 device you're describing. I love it.
00:50:20 Thank you. That's all I'm here to do is bring back the path.
00:50:22 I want to make one of those for the Galaxy Flip 5 or the Razr. I want to make a Klix that flips
00:50:27 around back and then flips up forward and overlaps the screen and gives you a physical QWERTY
00:50:32 keyboard that you can then flip away and put around back. It would sell at least 127 units,
00:50:38 119 would probably fail, we will never build it, but I would like to build it.
00:50:42 Why not have more mechanisms?
00:50:44 That's right.
00:50:45 It's just like, what if instead of a hinge, we had several hinges?
00:50:47 Exactly. How many parts can you make move? How many points of failure
00:50:51 can you introduce into this product?
00:50:53 Well, Michael, thank you. This is incredibly fun. I really appreciate it.
00:50:56 This was a great time. Thank you, David.
00:50:59 All right, we got to take a break and we'll be right back with a question from the Vergecast
00:51:02 hotline. Be right back.
00:51:04 All right, we're back. Let's get to the hotline. As always, the number is 866-VERGE-11 and the
00:51:18 email is vergecast@theverge.com. Send us all your questions, all your thoughts, all your feelings,
00:51:22 and we try to answer at least one on the show every week. First of all, again, thank you to
00:51:26 everybody who's calling in about TikTok and the Apple antitrust stuff. We're going to get back to
00:51:30 both of those a bunch in the next few weeks. So keep calling. Tell us everything you think we got
00:51:35 right. Tell us everything you think we got wrong. Tell us all of your feelings about all of it.
00:51:38 We want to hear everything. For now, we have a question from Michael.
00:51:42 This is Michael from Madison. So my parents are getting older and I'm looking to record
00:51:48 them sharing family stories so we have them for next generations. And I was looking for
00:51:55 advice on what to buy microphone-wise to record my parents talking like Ken Burns style,
00:52:01 like a cute couple. Any advice on what to do would be great. Thanks.
00:52:05 All right, first of all, I believe strongly that everyone should make Ken Burns documentaries about
00:52:09 their family. This is an extremely good idea. Second of all, Andrew Marino is here to help me
00:52:13 answer this question. Andrew, so much of you on the show today. This makes me very happy.
00:52:16 I know. What's up?
00:52:17 You have come prepared with what I assume is just a large slew of microphones.
00:52:22 This question gets more complicated the longer I think about it, which is why I'm very glad you're
00:52:26 here. So take us through it. What are your thoughts on this?
00:52:29 Okay. So I'm not sure what Michael is actually using to record. So I have a couple options
00:52:34 that will give a hybrid of options you can use. Right now, I'm actually using one of the
00:52:40 microphones, the Rode Wireless Me. Rode has like a slew of wireless microphones now that you could
00:52:48 just plug into your phone or plug into a camera or into an audio recorder. They are super easy to use.
00:52:55 They look a little square. You just kind of, oh, a big square.
00:52:59 It's a big square.
00:53:00 You clip onto your lapel.
00:53:02 These are the ones that have like the receiver you plug into your phone or computer or whatever,
00:53:07 and then the actual mic is what you put on whoever you're talking to, right? And then
00:53:11 they automatically connect to each other.
00:53:13 Yeah, exactly. So what I do a lot is I'll plug the receiver into my phone and I'll record a
00:53:20 voice memo or with Rode's recorder app or like take a video and it will take the audio from the
00:53:26 Rode Wireless for the audio input. That's a great option because it's very versatile if you want to
00:53:33 make video, if you want to be walking around, if your parents are like cooking dinner,
00:53:38 like moving around and all that.
00:53:40 Yeah, that's good.
00:53:41 The other one I have is the ATR2100X USB mic from Audio-Technica.
00:53:49 What I like about it is it's a USB microphone that's also an XLR microphone.
00:53:54 So if you want to plug into your laptop, into your tablet, into your phone, you can use this
00:54:00 microphone, but also you can just use it into any mobile audio recorder that has an XLR input or
00:54:05 your audio interface or anything. That's a great option. We send these mics to everyone on staff
00:54:11 for the Verge cast.
00:54:12 Yeah, this is the mic that I tell people, like if you don't want to spend
00:54:15 several hundred dollars on a pro podcasting mic, but you want 90% of the quality, I make people
00:54:21 buy this mic. Like I have recommended this mic to so many people who want to do podcasts.
00:54:26 Yeah, and they're very cheap. Right now I'm looking on B&H and it's $49.
00:54:31 Oh, wow.
00:54:32 Which is-
00:54:33 That's a really good deal.
00:54:34 Half the price of what they usually are.
00:54:36 Yeah, this thing is a good deal at $100, much less at $50. Okay.
00:54:40 I should also mention that the Rode Wireless packs range from $200 to $300,
00:54:45 depending what version you want to get. Those are a little more expensive.
00:54:49 So here's my question for you on this one, though. I feel like there's a version of this that is like
00:54:53 sit with your two parents, right? That's what you want to do. You want to have them sit on a couch
00:54:57 and talk to each other. Do you give them each a mic and plug it into a recorder or do you like
00:55:02 set the mic on a table and kind of point it between them?
00:55:05 Oh, that's a good question. So I actually did this with my friends recently. I interviewed
00:55:09 them at their kitchen table and I had two of the microphones, one for each person.
00:55:14 That's the ideal situation. That's like, I want the best recording. You can totally
00:55:19 just put table and have them talk into both of them, like a little press conference thing.
00:55:24 But since they're so cheap, like it's worth getting two,
00:55:28 maybe one for you, one for your guest kind of thing too.
00:55:32 Yeah. I mean, a hundred bucks, plug these two things into your computer. They each come with,
00:55:38 if memory serves, like a really crappy, but kind of useful little stand. So you could be kind of
00:55:43 all in for a hundred dollars on this project and have pretty solid audio.
00:55:47 Yeah. That's actually another reason why I recommend it is it comes with a stand.
00:55:51 It's not a good stand, but it is a stand and that counts for something.
00:55:54 All right. You said you had one more, right? What's the third option?
00:55:57 Okay. So this is the Sennheiser MKE 600. Right now it goes for about $350.
00:56:06 This is more of a professional microphone. I use this a lot for podcasting, field recordings,
00:56:12 or maybe I'm over someone's house and I'm interviewing them. I love the crispness of
00:56:18 this microphone. How do you think it sounds? It sounds really good. I can only see a little
00:56:23 bit of it on your screen, but this looks like a kind of proper shotgun mic.
00:56:28 Yes, exactly.
00:56:29 That like, this is the kind of thing that you stick on a boom arm and hold over somebody's head
00:56:33 on a film shoot. Like that's, this is a much more professional thing than either of the other two,
00:56:37 it seems.
00:56:37 Yeah. This is a great one for like indie filmmakers, even if you want to just put it
00:56:42 on top of the camera, but I love it just on a stand on a desk pointing right in front of the
00:56:48 person's mouth. It has just like a very detailed sound. If you want the Ken Burns documentary
00:56:54 style, then this is a great choice and it's affordable among the more professional microphones.
00:57:00 Yeah. The microphone world, especially in this realm of these like shotgun boom mics can get
00:57:06 really expensive really fast, but there's definitely, especially for our purposes here,
00:57:10 you hit the kind of diminishing returns plateau pretty fast. Like just hearing you now, this
00:57:15 sounds really good. I don't know that you would necessarily for like human archival purposes,
00:57:19 want something notably better than this.
00:57:22 Yeah, I think so. And this is just an XLR microphone, by the way. So you would need
00:57:27 either an audio interface or an audio recorder that has an XLR input, which is why I offer the
00:57:32 others.
00:57:33 Wait, I want you to do one test for me here really quickly though.
00:57:36 Yeah.
00:57:36 Just move around a little while you talk. Tell me, sorry, because this is the problem
00:57:40 with these shotgun mics, right? Is you have to be like sort of perfectly in front of it. And it
00:57:44 sounds amazing, but if you move, sometimes they get weird.
00:57:46 Yeah. So I'm talking to the left of it right now, which is off axis a little bit. You're
00:57:51 going to hear less of the clarity. I'm going back to the center and I'm going to the right,
00:57:56 go a little further back.
00:57:57 Okay.
00:57:58 Usually when they have this shotgun mic, you can go pretty far back because it can
00:58:02 catch your voice pretty far away.
00:58:05 Yeah. When you go back, you get quieter, but you're still very clear. But as soon as you
00:58:09 get off axis, like it starts veering into like I'm on AirPods kind of vibes that I'm listening to.
00:58:15 Yeah. So if you're interviewing two people, you're going to have to have them either switch
00:58:20 off between the microphone or you're going to have to boom it like a professional and move it around.
00:58:25 Yeah. It's always like you see the field producers for podcasts and whatever, and like half the job
00:58:31 is to just sort of shove the microphone in the face of whoever's talking.
00:58:34 Yeah.
00:58:34 And it sounds amazing, but it's a process. I'm torn between all of these. Actually,
00:58:39 this is a really funny three. As soon as you said the audio technical one, I was like, "Oh,
00:58:43 that's obviously going to be the answer." But I feel like the lapel mic is really useful for
00:58:48 somebody who is not used to talking into a mic in a way that it's just like it just sort of goes
00:58:52 with your body for somebody who is like moving around and isn't always focused on mic technique.
00:58:57 Having it just sort of live close to your mouth is always really useful. Then the audio technical
00:59:02 one is kind of a useful middle ground for not a lot of money. And then this one is clearly the
00:59:06 best sounding. Like it's actually notably better sounding than either of the other two, which I
00:59:09 didn't really expect to be the case.
00:59:11 Yeah. And you can keep going up. You can go to $10,000 if you want to get the best microphone.
00:59:16 But really like these ones are like very flexible in different environments. And that's why I chose
00:59:20 these.
00:59:21 Awesome. Well, Michael, I hope that helps. Please let us know which one you decide. And
00:59:26 we would love to hear the documentary when it's done. Andrew, thank you as always. That was
00:59:30 awesome.
00:59:30 No problem.
00:59:31 All right. That's it for The Verge cast today. Thanks to everybody who came on the show. And
00:59:36 thank you as always for listening. There's lots more on everything we talked about, including
00:59:39 that awesome video from our video team about noise cancellation, all this stuff about clicks.
00:59:44 We'll put it all in the show notes on theverge.com. But you know, read theverge.com.
00:59:48 It's a newsy time. There's a lot going on. Check it out. And as always, if you have thoughts,
00:59:53 feelings, questions, or other keyboards you want to see brought back to life, you can always email
00:59:58 us vergecasts@theverge.com or call the hotline 866-VERGE-11. We want to hear everything from
01:00:03 you. This show is produced by Andrew Marino, Liam James, and Will Poore. The Verge Cast is a
01:00:07 Verge production and part of the Vox Media Podcast Network. Neal, I, Alex, and I will be back on
01:00:11 Friday to talk about probably more antitrust stuff, more TikTok stuff, and everything else going on
01:00:16 in tech. We'll see you then. Rock and roll.
01:00:19 [MUSIC]

Recommended