how does the model work? to be honest, it is very difficult to say, this is the so-called black box. there are processes that, in principle, we do not know about. it is because of this that we were able to collect these samples of these flights. the model was already able to distinguish the future ones . all flights in such and such spectrograms where on the human eye it is no longer impossible to see, i.e. now we can talk about the fact that we are guaranteed by the probability of studying somewhere 50%, maybe 80 percent, we can clearly determine that this is the sound of a winged missiles and on the width of the front there is roughly 200 km and the depth is i don’t know, maybe now you will ask what the depth of detection is, we clearly know that a cruise missile is flying and this is the basis for this signal to be transmitted to the air defense system

for the adoption of an additional decision on destruction of this object, this is how it looks for cruise missiles and for airplanes. i'm afraid this percentage is higher than what you said because over these months we we managed to collect a powerful dataset from real flights in different weather conditions in different variations there on in different directions and it really works for cruise missiles, the second question is for sure i will not deceive anyone, the situation with drones is more complicated, that is, it is difficult to distinguish them from a moped, they are a difficult task, in principle, we believe that it can be done effectively, but unfortunately, no one since no one has a sufficient number of records, it is very difficult to get them, in principle we ask everyone we can reach to provide us with such records, but this is also a difficult

technical task, the records are made on mobile phones phones, they are very suitable, they are very distorted , they are cut in general, mobile phones are not suitable for training such models of artificial intelligence, they do not exist, they are designed for the human voice, the microphones in mobile phones are already designed for the transmission of the human voice, and nothing else . database for analysis, that is, an adequate database. that is, we can say that if we collected cruise missiles, this is exactly what ensures a high level of adequacy detecting a sound signal, you just go to the internet and download recordings or cruise missiles or even those drivers, well, it’s not very productive, it’s used, it’s called pre-training, that is, you can feed m.m. models of artificial intelligence, dear, such data will be useful but such percentages as us, but as

a result of the training of covered missiles, this is not a disk. how long did it take for artificial intelligence to learn to adequately distinguish a cruise missile policy in air space in general, for sure, yes, to say how long this task should last in terms of time, it’s probably not possible, it all depends on the specific conditions in which our team was, well, the main difficulties that i can convey to you are the main difficulties, such as if you have one sensor and it records the ether it is permissible. you have 24 hours of this ether of the environment around this sensor to find a flight in it. god forbid if this flight or god forbid that this does not happen, that these flights do not occur at all , but if in 24 hours this flight will be 10-15 seconds and a person has to listen to it a person has to hear it identify identify cut and save this was the main task of 40 sensors when we installed uh

hmm it was necessary to seat 40 people so that they listen to the air for 24 hours precisely due to this mathematical detection according to the level of signal activation we achieved that hmm, in general, the spans and many other things were already collected in automatic mode and it took us four months, according to your question , only after four months when all samples were collected in automatic mode, including far from using cruise missiles and manually sorting it, we were able to achieve the fact that in automatic mode the next samples were already collected for us. i carefully read your previous interview and saw some paradoxical things there, which were the biggest obstacles in the preparation of adequate databases were the bees and there are also other creatures that, relatively speaking, can pass close by the sound sensor, and the spectrum in which they are from sounds

is very close to, relatively speaking, after the flight of the rocket. i read it all, if so, how are you they got out of this trap. well, for sure, after all, the mooing of a cow cannot be said to be close to a cruise missile on the spectrometer. more probably , the reason was that in order to teach the model to ignore them, they need to be collected and said, this is it, this is the rocket. yes, this is them. it is necessary to find. well , it was very difficult to collect records of cow records, but when they happened, we had to listen to them and manually add them to the exclusions, in the same way, in principle , we did with mopeds and chainsaws, that is, in fact, the moped database by sounds now working hard to really distinguish a shahe from a moped, despite the fact that to the human ear it seems to be very similar, that is, but a nuance that should be introduced in if this problem has a very easy solution,

in fact, if you put the motor of the shahe alive in laboratory conditions, hmm, write it down with a normal number of microphones of sound studio equipment, it can be done uh, more than us, if there are colleagues participating in this project, they are engaged in exactly this kind of work, they advise us on microphones on precisely technical the composition of the e-e of exactly how it works, how the microphone works, how e-e, it all happens at this level and they achieve in their daily work, they achieve such results that they can determine the degree of wear of the rotor by the sound of the work, cool but this is not done in the conditions the fields are not in the conditions of the wind, rain is why our task is more difficult. and

when they talked about the fact that there are analogues, in what ways are these analogues from e-e close to your solutions and different in what is happening, let's say, conditional there e-e mental competition between projects that are directed on the detection of enemy objects precisely by the sound segment, well, it is very, very complicated, a very complicated topic, the main thing is to understand that there is probably no single correct approach to the identification and detection of air targets in particular. that is, we concentrated on our vision. how should it be done, the main approach - it is either we put many sensors all over ukraine and one of them will work with a high probability and during, for example, while the air target flies there for 20 km or 30 or 50, one of these sensors with a high with the probability of catching it or we do another approach

. that is, i am now telling you about other approaches that could be used. in general, we can absolutely act in the same way. it is a complex, like a large designer. we can abandon the antenna, then we will not take the directivity diagram that the antenna has we will have a 360° mirror, er, detection er, but it has its disadvantages, it has a smaller range, and such a microphone will not be protected from the external environment, that is, we have to create another receiving module, yes, we can do it, we can do it to implement, but having built such a network, we are afraid that we may help the state there for a few months and then we will have to hand over all the difficulties of maintaining such a system to someone. imagine, for example, we can, we see , in order to close the main lines of

flight of enemy missiles hm in the area of ​​600 of our stations is to cover the main main territories of ukraine along the perimeter and form a reliable field yes, that is, it is now 600 sensors for the whole of ukraine, yes, that is, that is how we see the entrance if put them at 600 and put them at one and a half two three four five thousand we will increase the density we will increase the density but who will service them who will be able to service such a number of sensors well, theoretically if we put these sensors on towers yes that is, in particular there on a telecommunications tower, then it is possible in principle, this is to create such a symbiosis where the emotional operator takes on himself , including the obligation to service these sound-receiving devices, and assign it to the local administrations of the military or others to it was such an integral project and your

specialists and e-e structures that take care of telecommunications are the same. it is possible to combine them in this way. our classic radar system for detecting air targets in general, the acoustic detection system is not much, it is already the main one, it is complementary, it should be complementary . on a system that only complements your basic radar classic, you understand why and that's why we need to keep a balance between the amount of resources we spend on this system and the advantages we get in our defense of our country, that is,

now we have an experimental area there with a width of 200 km and on the basis of which we are starting to work out new algorithms for detecting targets, and actually this is the set date, we created this uh, this , this milestone, first of all, in order to teach the artificial intelligence model to perform its work qualitatively and in fact, for you, there are startups where there are both mental and financial efforts, directly from your team, the participation of the state or the participation of local administrations. so far, there is no such thing. in the end, this is called the construction of such a system from which it will be possible to come with a folder of paper where it is described, show that it works, please take it, we will give you everything and implement it, because we still love military developments, they are made if, after all, from the side of the state, we spent on

this, you can say six months. we went and talked about money and brains. ours and also in principle and well, we do not call them competitors, mine just tried to explain that these are simply different approaches that can be used and they should complement each other, they do not even exclude each other from the map ukraine, well, now we are not at the stage where , by the way, the implementation of such new developments is significantly simplified, that is, the test cycle is now shortened, and the military told me that there is a new rule of thumb, if the product works, you take this product, you prove that it is really effective, and they take this one is for the provision of the armed forces, and on the basis of this , some conclusions are drawn up regarding the further fate of this or that project, as i

see it, in fact, with your project, we have come to the stage of pressuring the military, what are you guys? look it works. yes, we have come just for this stage . these tests will be carried out during the next month, a month and a half. in order for this system to be developed from scratch, taking into account our experience that we have had so far, all the problems that we have faced and overcome, we will see how it will show itself by the end of the year. and it is better now, how i understand that the main thing is that the person who

supported it, as i understand it, was like the leadership of the territorial defense, how can you say that in the alliance with the lviv military administration, it is clear that it can work. they trusted and you did it and it works now, that is, yes, local the authorities believed in this project, you can say from the first day, the city authorities did everything so that we could start working in that chaos, because the project was born almost immediately from the beginning of the war in the head, hmm, and after that the implementation already began already in april, the first stations were already standing, you didn't tell me the depth is 200 km - this is on our front. how far can we hear an enemy missile of the khas1 type, conditional, they say they are standing, as i said, they are standing by lines. each of the lines can have a detection range which we estimate from five

to seven kilometers, we have no way to order the flight of such a cruise missile in front of our sensor , ideally according to the directivity diagram in the center, so i hope that during these laboratory tests that will take place, we will be able to find out with accuracy up to a meter, what is our real range because they will take h101 and they will really provide real sounds that correspond to these rockets. so, i think that any jet engine would be enough for them. if the level of this noise is higher than the ambient noise level of the wind field of the road, well, to detect the sound , and if you have a sensor at a height of 10-15 m , then i am telling you that there is a risk, relatively speaking the reflection of a sound signal, that is, if we

simply detect the sound, it has no meaning, but if we want to know the direction of movement of the sound object, is it possible to implement it in your project as the fortune teller moves, first of all, you said for m-m for reflection, that is, it is also one of the main factors why we chose this approach to the implementation of such a project because we have a mirror and a parabolic antenna and we have a directional pattern not 360 a 100°. we can thus raise it to the horizon and we can cut it off what happens from the ground and the noise, for example, of the road, which is near our sensor, because we are essentially listening to the sky, and everything according to the directivity diagram of our mirror, it is weakened approximately by two times according to our calculations, that is, we can even improve

this indicator, that is, we can make a system that will listen only to the sky and it won't, it will also help to increase the detection range, and the second question, what is your second question about the direction of eh, this is often asked, that is, we we interacted a lot with those services and bodies that are involved in air defense and they always told us clearly that if we build a system with boundaries, then detecting the angle of the target is not a priority, because having two boundaries, when crossing one, the direction of the second is already clear. to the target is already formed on the camera, because we will already put it at an angle, he says that we can even determine this by trangulation, this is one of the possible approaches, and it also has its

disadvantages, and the difficulty of implementation if in our for example, two boundaries will be at a distance of more than 5-7 km, and it is difficult to place them very tightly, then one sensor will hear and detect this signal and by this time, when the next boundary hears it, the signal from the previous one will disappear already . thus inhalation is necessary. it will not be possible to do this . in principle, this can be solved if you put them closer, do not use mirrors, have 360° and it will work. well, again, the difficulty is in the density. we took a different approach. we have a group and a team of people who deal only with the problem determining the angle to the target, they do it in two planes, this is er, it means height and direction

horizontally and vertically, they achieve this at the expense of four microphones, four microphones that are er, 70 cm from each other, in general, this is classic, you can also already say classical implementation, that is, there are many in the system. that is, we take, say, another parabola in which there are four e-e parabolas, this is already without a mirror, it is four microphones that are located on cross-shaped poles, and there is a delay between the arrival of the signal between these microphones, then we go to triangulation, that’s how it works, it works, it’s a system that has many implementations. well, if you try to search for such projects in ukraine in google, you will even find many, and the americans have been using such systems for a very long time, well, other, other armies, that is, if we are there, by the way we see these foreign developments when there are acoustic sensors trying to detect a sniper's shot, yes, that is, and what are they trying to stuff or there are acoustic sensors placed on drones

on the wings of the wings in order to account for the difference in using microphones to detect from the air where the object that is shooting is located. that is, it is actually a technology close to this, which in principle is based on the presence of adequate software and a database for sound processing, so everything turns out correctly, there are two approaches, one is a microphone array. that is, there is more than 5 microphones, there are a lot of them and the calculation is due to this delay or a simpler implementation - it’s just four microphones on poles, uh, we are exactly that, the team is developing this system, it is working now it is not in such a complete form as our sensors, which have been installed and working for 7 months, do their job, but we can equip each of the sound stations with each of each sensor , we can equip such a subsystem for detecting the e-e angle to the target. such an

autonomous system, we have a module in the format of a parabola, some of which can be said to be placed on any object there at a certain angle and when it appears in the area of ​​sound vision of this, let's say, mirrors, the signal is simply automatically transmitted to to the command post, that is, to provide just such a sycentric network with completely autonomous modules, another approach is what you say, it can also be implemented, but unfortunately, in such different ones, it is more the so-called object protection , that is, there is some object that is needed to protect , for example, a substation and we place such mobile eh we can place such mobile eh sensors around it and today they stand eh on this object tomorrow they are transferred to others this is a different approach unfortunately we cannot disperse in order to develop all directions, but if we somehow managed to get more

funding in those developments, then this is also a problem that can be solved very quickly . help , who to look for, who to go to, we registered a public organization, it is called sound technologies, we will definitely start collecting in such an open form, donate to further deployment of this network earlier. we tried not to do this before. we will take it upon ourselves, in general, from the very beginning

, ukrzaliznytsia helped us a lot with donations, and also the public organization of persistent heroes of lviv region, they helped us financially at the very start of the project for so that here are these sensors that we already have to build. that's why there will now be a sound technology website, so you can watch our program now, i say cool guys, let's say yes, i want lviv region to be protected and, in principle, i believe that this project works. what about you? disappear, that is, there will be sites, it will be possible, in principle, to get acquainted with your initiatives, yes, yes, i think that it will be possible. so far, it is better, but i found yours through a publication in one of the publications that steals me , it is interesting because i see it. it is a really interesting thing, that is for now, you can definitely contact us through facebook, but this way we will register all social networks so that you can keep in touch with us

and also in order to see constant updates on all these functions that we just discussed with you, they will be developed will be added and are there similar projects in other countries that are implemented precisely with the principle of acoustic reconnaissance, well, in most of them, in most of them, what we saw when we investigated this issue, we saw exactly target angle detection that is, these are some kind of anti-drone systems because they are, in principle, commercial . they exist, we also communicate with them. by the way, we communicate because the bank is better. now, for the first time, ukraine is faced with such a dilemma that the number of enemy missile losses is high, it will be constant, and this means that the

defense system must be multi-level, multi-layered if there is no radar, then conditionally speaking, the question that the sound-metric complex seems to cost a lot is also not a question, it costs much cheaper than a network of relay stations, it is correct, in fact, that is, i think that in general the situation that ukraine is facing now has given many countries a chance to think about how they are building their defenses, these challenges are quite unique for us now and how do you see the development of the project now ? to interest the real military so that it is a state project and not just me telling you that it is an initiative project of a group of interested people, yes, without a doubt, this is exactly the direction we have drawn for ourselves now, we are keeping our fingers crossed for the test because we are absolutely confident in the system's efficiency and i hope that you

the test is they will only confirm. and you will be tested by the central scientific research institute of weapons, that is, he will give it, that is, who will conduct tests with e-e, those who tested it as a customer are e-e rep, that is, the command of the military forces of fish from the armor of ukraine, they are the customer, and the executor there, now we will only receive an invitation to participate in these tests , we will see who exactly will be the executor, we just keep you informed because the topic is called interesting, to say the least, if the direction of the perspective is, then it needs informational support because the military, despite everything we say, respect, sometimes there is a conservative stage and they will think that this well, it certainly can and should not be an approach during these seven months, we did not appear in public space well, of course, that is another question security, surely the more you talk about something, the more the enemy hears it, we tried very hard not to talk about it well, now the time has probably come, with

your help, we will also cover all the options . thank you very much for this interesting conversation, we got interested in supporting this project, and i will remind our viewers that the broadcast espresso channels was maryan sulym, a participant in the sound and army project, a serviceman of the 25th separate territorial defense brigade of the city of lviv, and the city of lviv, in fact, and the program will continue with the chronicles of the information war with olga we have incredible, very good news about ten-year-old pasha derevyanka. he was found. the boy's relatives took him out of the occupied territory where he had been for a long time and now the child is finally safe. we are very happy that pasha was saved for now. this is of course the most important thing, but at the same time we understand how it is now . it is important to help

the boy psychologically, to support him to find the right words, because he had to experience, without exaggeration, sheer horror at the end of september pasha and his mother fleeing from hostilities they tried to evacuate the village of kupyansk nozlovi, but the motorcade in which they were traveling came under artillery fire, here is the actual video from the spot, the picture is really scary and here in these zhigulys, little pasha was driving with his mother. unfortunately, the boy's mother died on the spot , then the father told us the details of this terrible story pasha, he did not evacuate with his family that day and remained in kupyansk after all this happened, at first the husband did not know anything about the fate of his son, the child was considered missing, and then the father they sent me a video in which he recognized pasha, they gave me a link, they sent me a link to a video where there was a plot in which it was told about the column being shot

, and there i recognized my son on the video, here are fragments of this video, it really is pasha derevyanko, here he is in one of the hospitals of the occupied the boy from luhansk is alive, but unfortunately he received a shrapnel wound in the hand and the worst thing is that his mother was killed in front of his eyes. you alone, with whom you were driving , and my uncle, my mother killed the children . on the spot, i include yes, on the one hand, i am not happy that he is alive. yes, that is the most important thing, but i can imagine his condition, which child has never been beaten before the eyes of the mother, and

maybe the uncle, there are seven, seven or eight people from the whole column alive. and there were about 40. so then ten-year-old pasha was wounded in front of his eyes , and his mother was killed. he found himself alone in the occupied territory, far from his home and relatives. all this time, the boy's father was looking for all possible options to save his son . the situation was complicated by the fact that all of pasha's documents were lost during the evacuation. now we still can't. to tell all the details about how in the end it was possible to get the boy out of the occupied territory, let's just say that his grandmother came to pick him up, there were serious difficulties to pick up pasha, then they were allowed to leave for russia, and from there the boy and his grandmother went to one of the european countries to his aunt