 Now, the microphone should be working okay, I'm just checking myself, I can't hear myself I didn't hear myself before, which is a bit odd, I have to double check, okay, you know, why? Yeah, now it should be, it should be working, yeah, you need headphones to be connected to your phone, that helps quite a bit. Okay, it's really annoying to hear myself, there's about five, six second delay, so if I don't respond on time, it's because of the delay. So we have this web applications, sorry, this web application, this web application, new web applications, IEG, spectrum visualization, it doesn't actually display anything, I don't know why there is nowhere, debugging is true, nothing is being plotted, window size is not being taken into account, we're still live, yes, okay, so we have this code, let's start a new chat, because this one is getting tangled, tangled in, we supposedly have some NANDs somewhere, number of NANDs in that, aren't we printing it all spectrum, so if I turn all spectrum on, no NANDs anywhere, one place that I can see, a filtered, it's getting confused now, which makes me suspect it lost some context of the code, a filtered data, should be filtered data, check it again, this didn't work, that'll be the least, I'm pretty sure we're doing this, is that processing, I don't think there's any processing on the front end, I don't know what that's talking about, first we're having this error, but also all spectrum is not NAND, yeah, I'm converting this to least already, frequency time spectrum, yes, front end, so I actually have that line, okay let's start a new chat, so we back to GPT-4, just using default, well because the data analysis which used to be the code interpreter is trying to execute on the OpenAI server, which doesn't suit what we are doing here, so let's start with index HTML, we have those prompts, that's how we started, so they, well they're doing something, but don't get any results yet, just saving that, oops, so it needs the context HTML, and that's okay, let's do the JavaScript next, and then the Python code, okay that's enough, and should be the JavaScript, it's many four contexts, as we're trying to actually solve specific errors that we're getting, okay that's enough, and that's the Python code, we just let it run, so we're doing this tool here, still have the old version there, and that's how it used to look like before the big collapse of the back end, and here we have the video there of how it meant to work, yes basically just scroll through a file, it shows you the row EG, and that's spectrogram, should be really straight forward, actually kind of like this, can just save this as an image, do I go full screen, copy that for a sec, just trying to get an image for the interface, because we could actually feed it into some ads, you know why it's already showing ads on my videos, they are not monetized, the channel is not monetized, so in theory there should be no ads, but they are, just grab that print screen, it should look like this, and this is what the front end should look like, front end layout, because we have different colors now, so get to this in a second, well that's okay I know what it does, HTML structure, yes, okay first of all, let's do one thing at a time, so we gave it all the code, please take all my code, well it's open source right, so the advantage of doing open source project, the fact that you can share, can align 78, it should be a quick fbix, restarting with start, run this, ooh, yes we have all this dance, everything else is working okay, let's go here for sec, so we're not, obviously not ready to deploy, have to solve the NAN issue first, this pretty much all of it is NANs, then we're getting this error from the browser, MI team NANs, shit, there indicates that your filtering process is producing NANs, not a number value, let's break down the issue filtering process to use the butter, so that's butter worth filter, bandpass filtering can sometimes result in non-value, especially if the data some has some extreme values, so I probably need to detrain that first, I would say, boundary effects, please boundary effects, especially when using functions like field-field, which is a forward-backward-filter steps, boundary data, replace NANs before sending to just, well that won't help because they're essentially all NANs, so the filter is not working correctly, would not work because we're essentially getting all NANs, doesn't mean the filter is not working correctly, shall we do the trending first before any filtering, so I forgot to click that button so the transcription, the voice to text will keep going, yeah this will only mask the problem, else let's just get all serious, the trending many can help, signal has linear trend, it does have a linear trend, this might affect the filter performance, move the trending process, right so we detrained first, let's see, a filter data detrained on, we want this before any filtering is applied, but it is, yeah the filter should be some very basic thing subtracting the mean of the signal to remove any DC, but that's what the trending should do to begin with, we'll compare those two later, check filter stability, if the number of poles above one or equal to one, absolute poles, don't have poles, number of roots, yeah essentially want to check how complicated this filter is or isn't the trending before filtering, pretty sure already placed the trending above code before you apply the filtering, I mean pretty sure it's already easy, let me temple check, so this is where field filters line 83, yeah signal detrenders line 59, so that should be fine, isn't the trending process ready before the filtering, should be yes, checking filter parameters is a good idea, so let's see if the filter is unstable, the four alpha is unstable, so why did you give me an unstable filter, should've had some basic filtering down sampling, 400 should be fairly low already, so why the numbers go 1, 2, 3 and then 1 again, so regards down sampling, fairly low, low pass, high pass sequential, that's a good idea, what we're currently using, should've given a default filter order, what is the filter order, it's taken from, yeah that's super large, okay so that filter order was too large, right, so now we're getting something, that was a simple fix, didn't actually have to change anything in the code, we need to obviously limit the filter order, this is actually the actual signal meant to be the bands and frequency, it's meant to be frequency, looks like the raw signal, well it looks like the deconstructed signal, so I know what is it doing, sorry it's breaking the signal into frequency, but this is still time domain, this shouldn't be time domain, and obviously here we should be able to see something, actually the seizure, so essentially it looks like the seizure broken down, I don't even know what's wrong with this whole spectrum, so the whole spectrum is that gray, it's not actually there, okay first we need to limit the filter order window size, it doesn't change anything, why, okay channel number, okay to seem to work, so we need to make sure, why is it a problem, by the way, what is it, what is the limit, so it's like something fairly low, yeah there will be a lot of work, what if it's the trend at first, can it then go higher, no, okay what's the first, this chart is called raw signal, so why am I not seeing the raw signal, okay, no it's not only able to take too long to process, it will be, otherwise it will be too sharp and the signal will be distorted and become unstable, stable, which we currently, which we are monitoring for, okay the other thing is that what's called raw, it's meant to be the original channel, channel number, the trend, whole spectrum, frequency scale, filter order, raw signal, this is the ID, new plot, raw signal data, signal is empty when the page is loaded and otherwise, yeah there's some good suggestions there, as we'll get to them in a sec when we actually see they should be able to display the raw signal, it should be a simple task, data equal data, data is data, am I using that time, no, this short time is not being used, a window size is not taking, it's not changing the charts, it's not showing signal, raw signal, it's the ID, raw signal data, no, raw signal data, filtered data, need to make sure copilot is not suggesting anything, add all raw signal to console, cannot do slice, I can just use print, right, this JavaScript not Python, so I can't pray, I can only print to console with label, I'm not using slice anymore, refresh the page, control F, it always goes to that launch, filter order, filter data, you know just sum up quickly, did quite a bit of progress, mainly with chgpt4 was able to generate some working code, github copilot was not able to generate a fully working code, but yes it could be useful to explain particular bits, it shouldn't really have the context for all the other files, in reality I'm not sure if it does, troubleshooting in console kind of helps, but primarily we're struggling with the structure of the data, so it's all there, but I think the bot is getting a bit confused, regards how the data is being stored might start a new chat that probably will help quite a bit, can still have four minutes, can try it now, let's do it quickly, see if it's able to resolve this particular problem, pop in html, JavaScript, the Python code, specify what the problem is, don't we, first of all, first of all, have this filter alpha unstable, we know how to fix it, we just need to change to reduce fault order, let's ask filter order, well we can do it, meanmax make it 4, just close the page, right, so this is 4, so we solve that quickly, single chat is not trying to signal, not thinking, yeah, my alpha band probably going up at the moment, considering I've been string for over an hour, it must be in a non-thinking state, so I might just continue later, see you in a bit, bye.