 So, Digikey Electronics, super partner, super friend of Adafruit, brings you on MPI. Lady Aida, what is this week's I on MPI? Okay, my I is on this MPI this week. We've got an NXP chip, and this is an interesting, let me check out this cool video. This chip is called the RT106L, and this is the little rendering of it. It's a BGA chip. And you might be wondering, hey, that sounded a lot like the RT1060, which is a chip that you've talked about before. This is a high-powered Cortex-M7 running at 600 MHz. So it's like, you know, basically got the speed almost of a Raspberry Pi or small Linux computer, but it doesn't have, you know, a full operating system in Linux on the complexity of RAM requirements. It actually has a flash and RAM built into it enough to run many projects. So it's like an ultra-high-powerful mic controller. It's really fascinating. And what I thought was neat is, I've never seen this before, but NXP said, okay, we got this incredibly powerful chip, and it's got all of these peripherals in it. Yeah. Check this out. We got like the CPU, we got the system, we got connectivity, STIO, Canvas, DMA, I2S, I2S, whatever. It's very powerful. Again, it's almost Linux power, capability, but in a mic controller. And I think they were interested in all this edge technology that people were talking about, like facial recognition, voice recognition. And they're like, well, a lot of these things, you can't do it on a small mic controller well. And if you do it on a microcomputer, it gets very expensive. Like if you look at something like the Echo Dot or something, it has a full, you know, Linux computer type thing inside of it running software. So he said, what if we can do a lot of this kind of processing that normally require my computer on a mic controller and have basically this edge-ready set of chips? So these, there's three versions of these chips. And you can see what you're getting is the same hardware as the RT-1060. So all the code and examples and the data sheets and, you know, valve boards and all that, that's the same. When you buy the 1060L, 1060A or 1060F, you get this machine learning inference engine, audio front end, video front end, all the codecs, all the beamforming for audio, the stuff that's so time-consuming, you get that with the chip as like a licensed software bundle. So interesting, same hardware, but when you buy these chips, you get a license to use their software. So there's three versions, the L, A and F. I had to pick one, so I picked the L. The L is for local audio. So this is a, again, this high-power chip, which has an inference engine that can recognize local speech. So there's no Wi-Fi connectivity needed to do some basic voice recognition. Now you can't recognize any word, you have to pre-program it with the words. And we've covered this in other videos on machine learning, how you train a model. And so the local voice solution, it's a little valve board, it's up to expensive, you can get it on Digi-Key, and of course it has everything. And the example code tutorial that comes with it is really great. And what's neat is how simple it is, you know, again, you don't need, if you want external flash, great, but for the most part, you can have the main chip, it's got tons of IO connected up to the I2S for the audio output, I2S to the two microphones for audio input, and you basically have a voice recognition and control system, and you still have that 67 megahertz CPU to do everything else you want with. There's also the A version, the 1060A, and the 1060A is the Alexa control, so that one has a license to use. So you actually make a little Alexa board, which is really nifty, I want to get one of those as well. For that, of course, you do need Wi-Fi because Alexa does, the wake word is local. This really is an MPI, this is really a new product introduction, that's like super cutting edge. Yeah, well there's finally available, they announced it, but it wasn't available. And then this one is the F, this is the facial recognition one, so there's a camera capability in this chip, and you connect up a camera and it can do voice recognition, facial recognition, I think it can definitely recognize A face, and I think you can train it by showing somebody the camera your face multiple times, and it can probably retrain to recognize between a couple different people as well. That's also got a PIR sensor, so there's basically three different valve boards, and they also have some videos from NXP. Yeah, we're going to play this back to back, they're about a minute each, so check them out. Hi, I'm Steve from NXP, and today I want to introduce our newest voice control solution based on the RT-106L for local commands. So this builds on our popular AVS microcontroller solution, but now we've enabled this to work completely offline, so no cloud connectivity required. This enables increased user privacy, lower latency from speech reaction, as well as potential to reduce the overall system cost. I'm going to give a quick demonstration here of the capability. Hey, NXP, window up. So with a simple custom wake word, which we can provide for any development, we're able to demonstrate in this case local commands simulating a window blinds being raised. Hey, NXP, window down. And it's that simple, and now we're going to show you how that works from Farfield as well. Hey, NXP, window up. Hey, NXP, window down. Alright, thank you for watching. To learn more about our new local voice control solution, please visit us at nxp.com slash mcu-local-voice. Hi, I'm here to introduce you to the MCU based local voice control solution from NXP. It features a turnkey and cost effective local voice control solution for products across home, commercial, and industrial applications. It includes production ready hardware and software that enables OEMs to rapidly and inexpensively incorporate voice control into a diverse range of products. This solution features the i.mx RT-106L crossover processor, complete with an audio processing front end and automatic speech recognition engine. In addition to creating the easiest and quickest path to adding local voice capabilities to MCU based devices, this solution eliminates the need for cloud connectivity and offers complete privacy with low latency. To learn more, go to nxp.com slash mcu-local-voice. Thanks. So actually, let's go to the overhacks. I want to show off this board and then we'll show you to get it. So this is the local voice control. And again, one of the things I like about the videos and what they talked about is this is great for low power and for privacy, where not only do you save money by not needing a wireless connection, but you also don't have to worry about customer support where you're dealing with wireless, which is always kind of the pain. And you don't have to worry about people thinking, you know, is my toaster literally sending my voice to the cloud and listening it on me. It's not possible when you don't have a Wi-Fi chipset, when you don't have any network connectivity, because you're doing fully local voice recognition. So this is a dual layer board, sorry, two PCB stack up. Underneath is going to be that main chip and probably some crystals and whatever passives it needs. And then it connects up to this mezzanine board, which has three microphones. I think this is either the audio amp or the Wi-Fi. And then there's a little speaker here as well. This is like a 3D printed or molded. I don't know exactly how this enclosure was made. It's all powered over USB-C. And then you can program this using an XP's IDE, which works great for this chip. And they've got example codes and a tutorial to take you through it. And I thought I'd also show the chip off itself. So it comes as a BGA, but it's a 0.65 millimeter BGA. So you actually have a shot for makers to make a PCB and fan out all of GPIO. It's really challenging when it's 0.4, 0.5, but 0.65, that's not too bad, I think. I'm going to handle that. So if you would like to pick this up. Where do you go? You go to Digikey for all the MPI? Yes. And the SLN local IoT, which is the local IoT demo for the RT106L. And then if you were more interested in the Alexa version, check out the SLN, I think, Dash Alexa 10-6A. And if you're interested in the face recognition demo, check out the RT106F. And that is this week's IonMPI. IonMPI.