 You're muted Michael. Ha look at that. Today's January 6th. Here we are. I'm still Being a lot of stuff there Hey, welcome everybody today is January 6th you all know that It's a lot of fun out there in the world But hopefully we've got good news here. I heard I heard a little bird and say that we had good news So all those clowns will be back in session in an hour Well on the on the more local front Who do we want to go first Chris there Chris B wants to go first sweet. Let's do it All right, so Cheers The beers were opened early today In conjunction with mr. Mr. Smith. I believe we have fixed the microphone issue Yeah, I I changed a C program which I haven't done and I'm not even sure how long but But I did import import audio basically what we did was We put a I found a patch online for port audio that takes the return code that was causing the failure and It skirts around it using something called an extra run that I still don't know what that is, but it's working My microphone has been running now for a half an hour so and Every time it gets that that error code. It just Restarts the audio and keeps going. So I think we're in good shape It's not the cleanest of fixes, but it is what we have And I will post my fix to the panticore dev channel after this meeting and hopefully will be will include this the other thing we found today is that There's a lot of defaults in not only our code but in the speech recognition code that are not suitable to our microphone So we actually went into some to some library code and changed it as well so The defaults are for 16-bit rates but we have a 32-bit rate not only a 32-bit rate, but they're a float 32 bit rate So it took us a lot of thinking that out. So that's in there and I also We also changed the sample Sample rate to match the microphone. So And the result is that the failure that was occurring frequently now occurs with much less frequency and when it does occur the We have a patch that it gets it gets around So yes, I think we're in great shape with the microphone so That was my day today The The underlying data piece of data the underlying data set that we use for training is that data set in the same at the same bit rate with the same Bit that's a good question. I Do not know the answer is no they're primarily 16 kilohertz Yeah, I thought there were 16 kilohertz like 16 bit mono or something. I mean they weren't they weren't particularly sharp That's because they were coming from an old pie But now you have a high-end Audio studio level hat delivering your your audio so Sample we can down sample from 30 from 32 bit to 16 bit and it'll be a lot Cleaner than if we up sampled from 16 or 8 anything. So I don't believe it's gonna be an issue in training Because I believe it's converted anyway in the training code, but I can verify that but There is code like this and the precise Chris Gusselin pointed out to we probably need to look at it as well, right? Correct. Correct. That's gonna need some Addressing in the runner. There's some hard-coded values there But but Josh the answer is we're better off with higher fidelity samples and worst case We can run them through a down sampling algorithm and bring that like my question was primarily it can our exist it Can we improve our existing model by improving the fidelity of the data set and then running it again? It's another training another training it through another training cycle Perhaps and the other concern obviously is when you train against 16k samples and you get 40k 48k input How well is that going to match up? So those are all TBD? but you know today is a Good day to celebrate Good awesome. Yeah, that's great. That's great news. So we're able to move Move into getting the rest of the demo working Yes, so tomorrow questions on it Can we um early and we and we do we understand you said something about like, you know I don't really know what that next next run or whatever the patch thing was can we We need to do some you know basic checks about where that code coming from and what it's doing and and all that sort of stuff And the other one just want to make sure are we changing hard-coded values in the In you know precise runner and stuff or are we dynamically loading those now from the configuration of it? So there's two answers to that first is As chris mentioned We had to go into some modules some python modules And hard code some stuff change some hard-coded stuff To match our hardware and we're not sure that that's absolutely necessary So the answer to your question is we just don't know we're gonna need another day to test with it run with it figure out the minimalistic changes necessary to get where we're at And by the way, this is a workaround. It's a band-aid because we know we're still getting overruns And it'd be good to make them go away Since we weren't getting them before But that being said I implemented is actually getting pulled into like debbie into and stuff too. So Yeah, that being said People with this problem far from it actually right. This is a valid workaround for now Until we can address those issues at a deeper level um, you know, it remains to be seen What it'll ultimately um turn out during our recognition when we run it But it was good progress. We're not done yet. We're not done yet, but it's good progress And we have a little bit more to go to get it over the finish line There's a change in the band-aid to core and to some configs. I mean, there's what I have on my device is all hacked right now I do know that he's not gonna be hacked and and yeah, that's that's tomorrow's work probably Yeah, yeah, we have a true monkey patch on our hands. Uh, so yeah, I mean We don't know yet chris. Uh, the the patch comes from a debbie in upstream So we feel pretty confident that if debbie and incorporated it It's not going to cause any troubles And the the x-run chris was talking about is Basically anytime you get an unrecoverable or what they consider a uh overrun or underrun error That's what the x-run stands for x is under or over And it's an over or underrun error. They used to just um assert out What this does is ignore the assert Reset the ulcer device and continue along its merry way So, you know less than ideal But better than where we were yesterday and something we can probably go to market with initially Awesome. Yeah, he said I'm very happy. I would crack a beer if it wasn't 8 a.m All right, well that that is very good news and but like everyone's noted it does seem like it's a patch. So I've taken a Uh a ticket in Jira and assigned it to ken. You can add details and reassign it as appropriate I just don't want to track of it. We don't have to you know, we can prioritize it as necessary But I just didn't want to lose track of the issue Um, great. So, uh, ken, do you have anything to add on your side or other things that are? No, no, no, I just held chris's beer while he solved the problem All right, sounds good. Well, I'm glad you stuck with it then Uh, can point at me in the right direction Uh, great. Well, let's go over to giz then uh Yeah, so I was doing some investigation on this branch as well. Um Uh Have set up a meeting with panicle for Thursday not friday because your friday makes it like, you know, just before midnight on friday night in europe, which is a pretty brutal thing um And we should we get this in then we should have um We should we should very much be able to do a new demo um to show the progress uh I did some investigation around The wi-fi set up um and tapping into xdg across containers um in pentcore uh, which it seemed possible, but I And chatting with them about yeah, the best ways to do that. Um Uh Particularly because it's on the system debuff. So we don't want to sort of, you know, just expose that to everything under the sun um But in the interim, I'm just gonna Get the rest of it set up so that it's it's in the image It's getting pulled into the image in terms of the demo and and it might just be on a timer to start with until we until we get some Triggering off the debuff messages properly. Um, but at least we'll be able to see it working and um Can kind of move forward on that Um other than that, uh There's uh, uh, I've been a really cool project coming from the community called magill to magill You know the um, what's the name from star trek? uh, anyway, uh It's a it's a front end framework. Um They've they've hooked up They've hooked into the the message of us and doing lots of stuff with amazon video and and youtube and and uh and codie and and a lot sort of stuff, um on a desktop so Uh, it's a cool little project. I just wanted to give it a bit of a shout out. Um I'll post a link in the team chat as well. Um Yeah, so for today we I'm gonna get that Wi-Fi stuff in I've been working on some voicom stability things We're seeing still seeing errors like randomly pop up. Um, part of that is because we haven't done a releases core in a long time So the chains of bug fixes that we have been doing are actually being run in our voicom thing, um But the market is really the priority at the moment. So Um, but as soon as we can get back to doing a release, that will be good. Um, 2102 is coming up pretty quickly anyway. So I think that what we'll end up doing is doing one point release In 2008, um with just all of the bug fixes and then do any of the Breaking changes for 2102 That is me. Okay. That sounds great. Um, Derek Hey all See so today we uh, we had a little meeting with, um Joe our uh similar down in Florida and Michael might talk a little bit more about that, but some of the things I was doing kind of um to help Johnny mostly get prepared for stepping in and helping with that is putting together finishing the spec documents for all the purchase parts and Getting everything ready to label all of that stuff and Kind of pass that knowledge on to me so that he can assist in the You know the handoff to Joe So, you know, like a lot of those spec sheets we didn't have for the different parts and stuff. So gathering together all the information that's necessary to understand what they are communicate that um I also had to make a change to our laser cut part This morning. There was some issues with the file. So fix that And uh, I got the audio chopped up for the um For the new onboarding walkthrough stuff. So At some point I'll have to handle it after I assume to guess to So we might need to have a meeting soon to Talk a little bit about that Or I can just send an email And uh Tomorrow I uh, so I got received the two boards from Kevin and um I would love to fire these up tomorrow with working mites so Um, let me know if you guys get that into something that I can get my hands on too And uh I'll throw them into some laser cut enclosures and we'll have two more How long it takes the panic for guys to integrate some of this stuff Sure, sure Yeah, that's what that's what I'm up to Okay, well theoretically, uh, if you push your changes up to the repo, don't they get them by, you know midnight One of the changes is something they're gonna have to apply the change to the board audio Something they're gonna have to do themselves Anything we do a core which we'll do tomorrow. Um, yes, you're correct. Okay. Got it um All right If we can if we can get all the details of that to like, you know down this evening Uh, then they'll have all of their tomorrow to work on that and Yeah, if it's just a patch to port audio, it should be pretty quick I can't see any if you like Getting in getting any changes in in a single day Yeah, I'll I'll write a post to the panic or dev channel. Um after this meeting Wait Uh, so Derek just to clarify. Did you only receive two of the sj 201 boards? Yes, I've got one for chris and one for myself and then kevin Sent uh one to ken directly which hopefully he got And then he sent one to panticore directly um Yeah, so that's where that's where I think he had more He could put more together, but that's that's the amount that he was able to get done Right, right. Okay. Yeah, I was expecting two more. So um I think uh, he's trying to deal one one came to me So you've got it No, I think no, I don't have it yet. I've got a tracking number though. Right. Yeah, I was All right, great Okay, I was expecting another one for rollover, but we'll uh I'll check in with him. He was having some trouble with the panticore shipping and customs and some nonsense. So um I'll check in with him after this Um Josh do you have anything you want to add? Nothing on the development side Right Okay. Yeah, likewise. I mean the meeting Derek alluded to Was just to check in with our fulfillment partner Crossing the t's dot in the eye is making sure we've got everything lined up So we can get the units out expeditiously once all the parts have arrived from the various suppliers Um, so that's looking good and um Yeah, that's about that's about all the news that is there so um Anything uh, people want to bring up outside of status updates or world news Yeah, okay Excellent. All right. Well, um I guess we'll call it there for today and we'll chat again tomorrow