THD 80 AUDIOSCENIC Virtual Headphones Using Headtracking and Advanced DSP

Show video

foreign ER again Hong Kong with yet another episode of  the THD podcast our first one for 2023 and it's   episode 80 we're clicking up there quickly and uh  let's not forget about our sponsor the ulti the   audio loudspeaker Technology International uh  great Organization for networking within audio   manufacturers so without delay let's say hello to  Simon in Japan good afternoon Simon afternoon how   you doing and good morning Mr Marcos Simone  a CTO and co-founder of audio Scenic based   in Southampton in the UK how are you doing uh  Margo sorry good morning I'm doing well and I'm   very happy to be here talking to you today okay so  yeah we we saw audio Scenic with their big press   release announcing their technology inside the  Razer gaming Soundbar product uh using optical   sensor to to targets and and head track uh people  and then do some audio DSP technology to create a   bubble of audio around people's heads so uh maybe  we'll start with a presentation to introduce the   company and the technology so yeah first of  all you can see now it is companion technology   so yeah so I'm going to tell you a bit what  is solution as a company is funded around 2017   um and it was funded by professor philippeci and  myself we were both working together at University   of Southampton in the in this place called  The Institute of some of vibration research   which is a a place in the UK a key Center for  a research in novio in Europe and in the whole   world and in this place in the isvr there was  a long tradition on working with um Crystal   cancellation systems reproduction of spatial  audio or binaural audio with speakers in front   of you okay and then we kind of built on that and  there was this long tradition coming from the 90s   there were like seminal papers key papers that had  Advantage technology and I'm Phillip or myself who   are working in this in in a UK research program  which was a big collaboration run between various   universities University of Salford University  of Surrey bbcr and B and Southampton and what   we wanted to to do there is that um basically the  whole the whole aim of the program was to bring   spatial sound into people's houses okay 3D audio  spatial sound immersive sound however you want   to call it because this is when the program  started 2015 the only reference back then of   special audio apart from headphones is 5.1 systems  multi-channel systems and these are systems that   um maybe some people can put in in some houses but  the normal people the common people don't put them   because they are not practical because you need  to put a lot of speakers all around you you know   because they're expensive they don't have space at  home and we said well can we can we give the same   experience or even better with the SS sound bar  and this is when they basically we started working   with loudspeaker arrays and we put a head tracking  because um there was I mean the technology of   Crystal cancellation already existed for many  years okay first but then goes to 1968. however   it had a very very strong sweet spot and this is  what basically we we solved okay if I advanced   in my presentation just to tell you a bit about  the company this is um audio Scenic right now at   the moment we are all based in in South Hampton  in the United Kingdom because we are associated   with a university where has been out we have all  our Engineering in Southampton and then we have   um a business development presence in in China  and Shenzhen and in Taiwan okay that's our company   right now we just closed another funding run over  series a just before Christmas we haven't even   done an announcement because we were busy with cs  and all that but yeah we expect that in the next   year we will grow to be twice twice the size and  be able to follow more commercial opportunities   this is uh what makes audiosynik a particular  company and a different company from others trying   to do like let's say virtualization effects  or post processing effects or three audio   with speakers and is that our technology there is  always a symbiosis an integration between um audio   and um head tracking which is done with computer  vision or with other methods and what hetracking   does is knowing the position of the ears of  the users with respect to the speakers okay   and then what we are doing is that we're using  some field control is the actual the technical   name we are directing How Sound is radiated in  space so that this is always the best at the   years of the users and because of that what we do  is that we can create large difference in pressure   between both years of the users for example and  create what we call virtual headphones all right   and this is really how in a nutshell in a very  simplified way this is what we are doing you   know for example with a sound bar we have two  beams one beam from the left from the left   ear Left Channel of a neural signal another  Beam for the right ear right channel signal   okay and then once we have that we have virtual  headphones and what happens that if the user moves   then what happens is that we adapt these vertical  happens in space right and then they always have   the best spatial signal line because they have  virtual helpers then we can render any content   okay like any special of your content this is  like a simplified version for one person this   is a research that we're currently doing into  how to adapt this for for more than one person   you know so I have layer in my bag very large  May array would use that as reference designed   to to basically develop what we want to do which  is our home Cinema solution okay assistant Network   for one person for two people and then it can  turn into itself also like a more uh normal   sambar with all bianos or something like that  okay if we want to invite our friends to watch   the Champions League or the Super Bowl in America  or something like that you know so yeah basically   just just to to repeat again the good thing is  that our technology because is is in a nutshell   headphones so it is compatible with anything okay  from stereo 12 degrees audio to binaural and 3D   sound in the middle because any format that is  special you can convert it to binaural you know   and and the combination of a good content and  also like this way which is basically a creation   of virtual Health from these headphones  that you're not wearing gives a very good   um basically some experience or performance okay  and I just want to put this um this slide here   because this from the racer Leviathan V2 Pro which  is the first product with our technology that was   released at CES the sambar got 12 Awards of Best  in Show okay another Cinema award is likely expert   like a best technology in the show because see if  it's a new way of listening to spatial audience   like basically you're in front of the device  and you're hearing sound all around you and   because you are not wearing anything is very  very realistic and very immersive you know so   so yeah so basically reproduction of  binary with speakers very very very   it's very plausible I'm very realistic and I  think because we have been able to tailor that   with the user adaptive experience you're always  on The Sweet Spot it's very yeah from breaking   and in the end it worked very well okay what  what we do in audio Scenic as a as a company   is that we provide a set of um the core our core  IP Technologies the our 3D audio rendering okay   that we call it virtual and then what we Pro we  also provide in our software is a set of Audio   Dynamics and Equalization sometimes then you're  making a product you can have certain effects like   backwards like a VRC which is on the amplifiers  this can also live in our software okay sometimes   the customer can want to have that to release in  speed to Market and we also it's very important   something that we need is the head tracking  software okay and in this case the head tracking   company we have our own head tracking we can also  work with third-party head trackers okay would   you uh those guys like Toby out of Sweden would  you have a relationship with those kind of guys   and we can work with them yeah we haven't met them  openly from the moment engage with Toby but we are   happy you know yeah okay you know as long as any  any tracking that they provides the accuracy that   we need and the speed that we need is welcome  yeah yeah they do the the retina tracking   um I'm sorry I'm going very fast through those  lights if you want to ask me anything at any   moment please no don't worry we will we will okay  perfect so um yeah basically when we combine this   software that we have developed okay with um a  speaker array and speaker array is basically an   aggravation of speakers in in certain form this  speaker array can be a sound bar but it can also   be embedded on a TV or embedded on a smart speaker  embedded on a laptop okay so basically one one   set we combine these two things we have what we  call personalized listener adaptive 3D audio okay   and we say is next Generation user adaptive  spatial audio because it adapts to the user   okay so yeah and basically what we  do in audio Scenic is that we work   um we license our technology to to partner  okay that's our our business model actually   as a company yeah and now I wanted to  talk to you a bit about what use cases   where you can apply the technology okay  okay so the first one is uh basically   um Next Generation uh Soundbars for immersive  gaming okay and this is a renderer which is a   couple of years old but it's basically it says  Corsair doesn't have to do anything with courser   there is a one of the first iterations so when  we developed the actual Soundbar that they became   later the the Razer and Leviathan V2 Pro yeah  and the idea that you're in your desktop like   I am right now and I'm I'm listening to some  with that and I'm I'm playing video games and   I don't need a headphones because I'm listening  to my my colleagues my my colleagues my enemies   or my friends yeah get it or whatever and what  we've done is uh basically we've implemented   um these um December in two ways one and it's  a standalone soundbar which has an SOC inside   in this case an nxp imx8m and is this SOC will do  all the head tracking and we'll also do all them um audio VSP okay okay one way where we  have implemented the things we call that   the Standalone because Standalone devices  like it's like the Leviathan which you   probably you just take the Soundbar it works  you know and then a other method for the way   we were we have implemented this is uh with  um in Windows 10 Windows 11 as well in which   we have upload part of the processing into the  into the computer okay and in the end is like   two different implementations one of them  can achieve a lower bill of material okay   same the idea this is also something where we  are working at the moment uh basically and this   is our laptop reference design okay and this is  how to show how the technology can be applied   into laptops and basically this is something that  we build a mock-up with 3D printed but we built a   small somber with actual laptop drivers okay okay  yeah to show that the technology could be scaled   down because within that laptops is also a very  good place where the technology could be applied   because I mean a laptop and you have a camera  already there now so you don't need to put the   camera you can use the same camera that we're  using now for for the video conferencing that   to to do user tracking and you can embed speakers  you know and and does it does it create enough of   a privacy bubble where the the sound doesn't  spill out too much to the neighbors beside you   and no because I will I will tell you I will  tell you another slide later about privacy so   we are also working in in products that create  privacy okay but to create privacy you need more   speakers okay okay so to do a spatial sound  as I'm going to explain later to do binaural   reproduction Willow speakers two speakers  are enough okay if you really want to and   um for example for robust 3D sound reproduction  we recommend to use five speakers okay   if you really want to do beaming of sound you need  more or less speakers and more is the more the   merrier okay so we also built a system that has  been for main or for example this one you would   lower your YouTube channel you can see some  videos I've been forming with this array this   array has 24 speakers okay it works very well for  being for me but that's what you need right okay   uh beamforming and audio focusing either or  you use traditional speakers and traditional   array techniques with lots of loudspeakers or you  use an ultrasound ultrasonic arrays like other   technologies that we're providing we're using  ultrasound array and the ultrasound ultrasound   array is very directional the same previous  low frequency but it is free Direction but then   um directional audio um is some something or  like audio something where Size Matters and   in directional audio matters a lot the dimension  of the array is what the term and the number of   speakers about the dimension of the array is what  determines if the array is directional or not okay main thing so yeah with this we hope that we'll  see laptops coming within and incorporating this   technology in the in the years to come okay good  very much this is also a render of uh something   we like to see in the market also in one or two  years which is then it's a home Cinema somewhere   and it's this this uh animation that I showed  you before where somebody can transition between   one two people provide a specialty for two people  and then work as a normal summer right right okay so this is basically um where we want to use sound  focusing okay and we think that if we have some   bars that have a large enough number of speakers  let's say 12 or 16. we can use them to create a   bubble option and that that basically works we had  in in our suite at CS we had a demonstration of   this why is this um important because well many  people are are since copied we have many people   that are working at home and working in the office  you know what do we want to promote to do with   this we want to promote again a virtual headphone  communication and that's that I am in my office   and I'm receiving a call and my colleague is there  and my colleague is not the Star by my call okay   right now we are doing that with with headphones  but you know headphones can be very tiring at   the end of the day you can see people working on  call center working on this place if they could   have that communication without headphones  it will be much more natural you know right   apparent because we are basically creating  headphones we can render 3D sound and what   we can do here is that we have a row of people in  a meeting and when they speak we can acoustically   locate them where they are you know and there are  some teleconferencing clients that already have   applied this technology and adopt this technology  like um dolby.io for example has this technology

I don't know if they have already implemented  positional audio but um teams and zoom they say   that they were going to adopt this technology  so it's also that you can do something that you   also can do with with our technology with using  virtual headphones and bubbles for specialty   reproduction okay okay and for teleconferencing  and the other um use case all right where we are   working and where we have applied our technology  not nothing commercially but something we would   like to see and then next year to come this  Automotive and last year we did a project of   um in this case it was a independent listening  zones okay where basically we were having one   every one person and that was I listened to music  and another person was listening to speech which   in this case was the EPS okay with the driver and  apart from that because we use our technology and   again we're tracking the position of users inside  the car we have um sound which is independent   between two people and that cannot be spatial all  right so I mean this Technologies and many people   are trying to do the same things in the whole  world we need to take them with a pinch of salt   you are able to do um one person has speech  the other person has music some you know   reproducing two different uh contents of  music with a kind of quality is very it's   almost impossible here unless you have ridiculous  Arrangement Arrangements of speakers here and you   know like that we are thinking on the car cabin  of today is quite hard you know right very much   and then also to show you um something which  is a our tuning tool okay that is basically   um something that um allows us to um control the  quality and control the AQ and control the DRC   and control the spatial characteristics of our um  algorithms okay and then the acoustic engineer can   basically play with that to give out a certain  color or a certain tone to their to the product   okay so we provide yeah and this is basically  the the um kind of company presentation and   Company pitch I now would like to dive a bit  into the theory of how all of this works you   know for if somebody listens until here and they  haven't fallen asleep now I would understand that   we'll send them to sleep finally you know which is  the theory but it's also very fascinating really   it's a good thing yeah how The Thing Really  Works and then it's basically all to do about   um free sounds and and and basically reproduce  sound so when we have normally a 3D sound we   use a binaural microphones okay which you  use a dummy head or some kind of a we can   also synthesize it via hrtf but when we get when  we are thinking about 3D sound for headphones we   get two channels okay we will get our Left Channel  put my pointer here so that you can see me better   we'll get the Left Channel and we will get the  right panel here okay and then if we capture that   with our enroll ahead what we do is that then we  basically feed it straight and reproduce it with   headphones you know and I think um most people  probably is familiar with this through Apple music   or maybe they have listened sometime in their  life today binaural Barbershop which is very very   very is a very good piece of content so basically  they record it with a dummy head and then you just   listen with headphones and you have the feeling of  being there and that's the power of binora which   is is so good and so captivating so what we want  to do is well we don't want to use headphones okay   we want to use virtual headphones and for that  what we are going to do is that we are going to   create and use a DSP system that is going to  basically create this virtual headphones here   okay it's going to produce a maximum of pressure  in one side and a minimum of pressure on on the   other side and it's going to give us the virtual  right Channel and then a maximum of pressure on   the left T are in a minimum pressure on the right  theorem will give us the virtual let's summon and   then anything that we have recorded here then  we're going to reproduce it and then the user   is going to receive here the information as if  sources were coming from different positions and   he or she will believe that is there acoustically  and that's really how how the technology works   okay um basically how can we do that okay so this  is the amazing laser pointer I want to see it   actually work yeah so how does that work  so let's say basically I have two speakers   um in space and we are going to be talking  here with two speakers at the beginning because   um it's easier to explain but um and then I  have my speakers connected to a computer which   is basically a digital signal processor right  and for for these animations I want to thank   to philippasi microfounder because he created  long time ago for for some of his lectures and   he'll he let me use them here okay so we have a  focus a point and basically what we know here is   we know in Space the transfer function between  these speakers and this focusing point and we   know this transfer function because we have  characterized it because we have measure it   or because we have modeled it you know because in  space sound is going to propagate proportional to   the speed of sound so I can know and I can model  how long one speaker the one that disclosure takes   to arrive to this Focus point and one speaker  the one that is farther away how much does   it take to arrive to this other point so when I  know this time means what I can say is like well   if I want if I want both of them to arrive in time  at this point what I'm going to do is that I'm   going to introduce a delay in the p in the area in  the speaker that is closer to this focusing point   so that I will delay it and then both of them  arrive here in time okay and then what happened   is this is this one has fire first and this  one has fired after and then a wave is going   to propagate that in this direction it's going to  create a uh playing waves if we are very far away   and it's going to arrive with maximum of energy  at this point okay and that's a constructive   interfere so my speaker has both gone and inputs  have played an inputs that is arriving in time   so with that we have what we call a constructive  interference okay it's what we have here so   that's great we can basically then say okay this  focusing point is my ear I want a maximum pressure but then what is happening if I want to do some  cancellation I want to create a destructive   preference what I'm going to do is that this I  know where I where I want my cancellation point to   be so what I will do is that well if I basically  change the face of one of them they are going to   arrive with cancellation of this plant okay and  what happened this guy outputs first because he's   farther away this guy also outputs but with  negative phase and you can see that in this   angular position the energy is very much reduced  in theory their energy is zero okay and basically   practice is something like that if you go to the  to the web page where the racer Leviathan V2 Pro   SM is a feature you can there is an explainer  that explains the informing technology and they   have a lot of a great video also explaining  the constructive and destructive interference   okay problem in this case is 2D in real  life it happens in 3D you know yeah we   can put the put the links to those in the  description so people can check it out later   on that that'll be very cool so yeah basically  this is what happens on a nutshell okay how the   technology how about is the the physical phenomena  how do we Implement 3D audio with speakers okay so   basically um what happens is that um well let's  say let's say I have this Source here this that   I want to to reproduce which is I'm sure that  basically is here at my right and that if I record   it I record its input response or if I record  a certain length of audio at both my ears it's   going to produce different content on my right  Channel and on my left my channel my left here   that will basically in this differences that  arrive to the two years will be embedded the   binaural signal that will tell my brain  that something is coming from there so   and we said normally people will  listen to this with with headphones but   what happens if we want to listen  to it with speakers okay sorry   and what we're going to do in order to reproduce  this with speakers is that we're going to use a   network of digital filters okay and it says here  digital filters for inversion or loudspeaker   response because what we call this inversion  because we want to invert the speaker response   we want to get rid of it and that's basically how  our technology works with you you have a sound bar   in front of you but you don't want to hear to the  number you want the summer to disappear because   the only thing that you want is the neural  queues that are arriving to your ear right   and again here we're here basically talking with  two law speakers because I mean stereo is such   a common uh setup and people have tried to use  that and to work with this for many many years   the first patent of this kind of technology is  from 1968 okay from atal and zeroider basically   they they've represent the first system so I  mean one thing I could do for example is just   try my stereo speakers and play the my normal  recording that I have and play my right channel   to the right speaker and play my left channel to  the through the left speaker all right and I can   do that on the right channel will be reproduced  by the right speaker what happened however that   I'm also going to here on my right ear the let's  Channel and that's going to mesh I'm going to   interfere with my binaural perception because  I really if I really want to hear binaural well   I need to hear only for this guy listen on each  ear okay right and this is crosstalk and that's   why this technique is in academic terms is called  in scientific terms called Coastal cancellation   you know so the Razer somebody says that this  3D audio being forming and it's true it's been   forming because um with when you have more than  two speakers you start to create beams which are   kind of pronounced but on on the back of that  what happened is that you are doing Crystal   cancellation as well you're trying to separate  what one channel is playing on the other okay   and basically we're going to achieve this by  using a filter okay that is going to cancel what   the whatever signal of one channel is arriving on  the other channel okay so is this is this kind of   I like the with the Beatles a lot of people love  to listen to The Beatles on headphones because   George Martin had recorded the the left and right  separately so a lot of people are always obsessed   with listening to The Beatles on headphones  rather than in a stereo speaker environment   uh yeah really um maybe you know I think so if  you listen to that for example that you're saying   that's a very hard punt um yeah stereo track  which is maybe like in the 70s 60s like that   if you hear it with a first floor cancellation  system like with a razor or somewhere you will   hear that things sound very very special to you  you know okay decent here and this one here if   you listen back with a normal stereo setup things  don't sound here things sound like in the speaker   and the other speaker you know okay um but uh  for for binaural audio reproduction you really   need to separate what arrives to one year into  the other ear you know and that's why basically   you can do a test you can listen basically with  the barbershop with with headphones and you can   basically then listen to it with stereo speaker  and you will see that with headphones it's you   are immersed in the in the in the barbershop with  the stereo speakers with the terrorist speakers   you know and I encourage everybody to to basically  buy the razor or somewhere because it's the first   system in the world that let you have they say  I mean it's not the first system in the world   but I would say the first item in the world that  works in a robust Manner and that is just turn it   on and works you know that is practical and you  can you will be able to yeah to listen to any   binaural content to play any video game or also to  listen to music in a way which is very it's it's   very special you know because you just put it and  you are here in Stereo to the other you know okay   and then they're just basically digging  a bit into the into the theory here   um so we need to use four filters okay and that's  very important and it's because this is my four   filters for foreign version and that's because  um if if I have one speaker only here and I want   to wake up equalize my um whatever I listen in a  given year one filter is okay is enough you know   however now we are asking two things to the to  the system with two speakers we are asking to have   an equalized signal in one ear and then cancel  whatever I'm listening in the other year okay   and in this case this is a system of two  equations that needs two variables to be   solved okay once and one equation is one of my  years the other question is the other of my ears   and when I'm asking to have maximum pressure  and when I asked him to have minimum pressure   and that's why why per Channel I need to  have two filters okay I need to have one   that repeat this well both of them work at  the same time they make sure that they have   maximum pressure in one year and a minimum  of pressure on the other area Okay so with   these two signal six with these two filters I  make sure that my right channel arrives to my   right ear and I make sure that the my right  channel doesn't arrive to my left here right   and then I do the same with the Left Channel I  make sure with other two filters on the left with   the lib that are basically connected to the left  input my left input signal arrives to my left ear   and then I make sure that my left input  signal does not arrive to my right ear okay   this is very cool in theory you say yeah it has  to work no it has to work in practice it's a super   complicated system okay and basically you say that  this is their Infinite combination of two filters   to get left ER scene all right okay and why it's  so complicated well so let's let's take a look at   how these filters that I need to do and what the  speaker signals look in time all right now this   is again I need to thank uh for to for this light  to somebody that was an isvr and takashitake that   basically also had a um another company that was  working with with 3D sound he created these lights   I think time ago and I think he's playing the  concept there very well so let's say that this   is my target binaural signal okay this is what I  want to listen on my left ear and it's in impulse   I only wanted to arrive to my left ear so but  let's see how we can make it so what we can   do is that my speaker my left speaker place  this impulse because she's the speaker that   is closer to my left ear in front of me okay  he's going to reproduce it I'm very good that   impulse will arrive to my left ear okay this is  what I wanted amazing perfect I have my inputs   however because sound propagating the space  everywhere and we could we cannot stop it   unfortunately these inputs that this has arrived  to my left ear all right a bit later to my right   ear no this is not what we wanted in our Target  binaural signal we only wanted these symbols here   not there because otherwise it's going to mess  with our perception of the spatial sound so the   algorithm because she's very clever and because it  knows that in the right ear this input is going to   be received says no no don't worry because before  it arrives I'm going to arrive and I'm going to   emit from the right speaker another impulse in  opposite phase that will arrive and will cancel   this one very good we have canceled the impulse  in the right here what happened what happens is   that these inputs that we have made to cancel this  inputs at the right here is then going to arrive   to the left here and that's not what we wanted  because we just wanted one impulse we didn't want   that this is a different content that the one that  we want to initially and then the algorithm says   again no no don't worry wait for me I'm going to  now make another impulse in opposite phase that   will cancel this one on the left here and you can  see there but unfortunately this Imports is going   to continue propagating and it's going to react  to the right here and then the algorithm that   is clever is going to emit another Imports in  opposite phase and you see for Yaz reproducing   this first inputs that we wanted in this speaker  all these other time information is required   you know and that that's why these  systems become so difficult to basically   um control you know and Implement because they  require a signal processing um system which is   kind of a very stable sometimes and that needs  to Output loss of cancellation energy this is   all energy in the end that is output and that's  put some constraints in the in the dynamic range   of the speakers you know and all of that this is  how it looks in time so just to play these impulse   we need to put all this garbage after okay and  that way they become so complicated if we look   at how this uh basically um how this looks in the  frequency domain it'll look something like that   and what happens is that um this is the the norm  of the of the crystal cancellation filters and at   some frequencies is needs much more energy than  not others you know and this is what we call   ill condition and this is a very common problem of  Crystal cancellation systems and in the end of any   any kind of a array processing line reprocessing  there is always like constraints in the dynamic   range that you cannot put you know and I'm  going to talk about a bit more later about that   so how does this look in the time domain okay  it looks at basically an assimilation just   to play one impulse look how much we need to  play this guy is going to receive an Impulse   on the on the left on the right here okay  see this first impulse is going to arrive   and he listens it but look how many others we need  to push so that nothing arrives after the first   impulse you know I'm not very that's really  very yeah it's very fascinating and very and   that's the really technical problem of of Crystal  constellation you know okay very much yep and I   have some four more slides that I can talk to you  about the more other than technical elections then   um uh when you're developing a system like this  is uh there must be a fairly lengthy process of   characterizing the acoustic system yeah yeah  basically so so there are a lots of uh really   um yeah you know like here so we I've been working  on this a lot on people before me have been   working a lot and and that's so many measurements  you know we have even basically this thing that   you see there is a robot for automatic measurement  of system you know it has two microphones and   basically we can move the microphone around  and we can say where where people are how   the sound really is you know because uh in  the end there are there are so many things   that affect this sound this is not only not only  the system also the room you know and that's why   that's why the head tracking becomes so important  here you know and I can go basically a bit forward   in some basically slides that I have here this is  how small the sweet spot is okay for those systems   and this is basically a simulation again just  using two loud speakers for for Simplicity and   it's the headphone effect and this is kind  of how the headphone effect looks with with   angular deviation we can get a this is DB for to  put the label here but this is DB of cancellation   when you are very much on where your filters  are doing this cancellation you obtained a   lot of pressure difference okay but if you move  away even a little bit okay where we calculated   our three degrees which is 1.5 centimeters 1.5  centimeters still is is kind of a okay but in the   moment you'll move three centimeters which is just  very small position you see there are some areas   where this constellation that before was kind  of large is now like really being really reduced   and yeah okay so um uh the uh filter  characteristics you uh fixed at design time   so you know what our technology does is that it  calculates um maybe 30 times filters per second   okay but you must be a layer where you've  uh characterized say the acoustic system   um calculate against so so our technology is based  on on models of physical models or propagation   and that's what we use for characterization you  know okay and we that's why we can we can very   fast and estimate the position of users you  know because um make people another foreign and in terms of uh frequency range or bandwidth  of how well it works I mean uh you uh can't really   have a huge array of very large speakers for low  frequency sound so you kind of taper it from some   into mid and high frequency and then low  frequency is kind of a uniform field yeah   so I think I basically um our technology at the  moment works as um so what we what we are doing   is that we provide a low pass filter and then  a high pass filter an open filter for base that   basically the base is not diagonal directional  but then everything above the high pass filter   okay will be controlled by the algorithm  you know a basically what I was going to say   the technology can be applied for any number  of speakers okay an arrays of any sizes you   can see here with from a laptop to the soundbar  to basically very large arrays that we have been   using here for development okay yeah and basically  um in this case for example on this large one that   we have here on the left hand side I think the  cutoff frequency for where the algorithm was   starting to work where 100 Hertz you know 100  Hertz it's still very low actually expected quite rapidly yeah so from our from my signal  processing point of view this thing needs lots of   signal processing you know yeah so I mean in the  Razer zombie has five speakers each speaker has an   independent Channel and basically has a filter  it's a network of filters yeah yep very much   okay very interesting and uh just uh so we've  talked about that uh aspect of it and the head   tracking is something a kind of critical element  um uh it's got something more you could tell us   a little bit about that because uh there'll be  another it depends on the hardware that you have   and things like that too I guess yeah so so I  think basically this is this slide where you   see there's a red line here that's a let's say  it's kind of the minimum that you should get to   to start having a a good specialization okay I  would immersive minimum Crystal cancellation if   you don't do head tracking you know and you move  even three centimeters there are some frequencies   where this you're not getting this you know okay  so the higher the better the better yeah yeah and   really you would like these two or even this is  this is what happens when you use two speakers   you would like this to go even even higher you  know yeah and that's why that's why we recommend   also to basically use more than two speakers  and all I want to show you what happens here   this is the crystal constellation of headphone  effect the higher the better okay so in this   basically left hand side we have the CTC headphone  effect and in this right hand side we have the AE   which is the array of Fort and basically it's a  it's a measure of the electrical power that the   system needs to obtain this headphone effect yeah  so this one on the left we want it to be the as   large as possible this one on the on the right  we want it to be as low as possible you know   because if we need more Norm it means that we are  basically compromising the dynamic rate range so   if we just have two speakers here which is this  is basically has been what the whole of your new   industry audio industry has been trying to do  with for many years just use two speakers to do   crystal cancellation what happens is that there  are certain frequencies where you naturally you   are not obtaining cancellation okay because it's  it's it has to do with a special biasing and there   is a given frequency where you cannot um have a  maximum and cancellation yeah wavelength related   essentially yeah basically that happens and this  is the effect that you see here yeah but if you   are the third speaker okay you basically increase  the specializing frequency and you only have a   minimum a much higher frequency so you open in the  range and that's why we say to to people who want   to implement our technology use three speakers a  minimum but use more than three speakers because   um you use four you increase okay your range and  if you use you use five you even increase it a   lot yeah and in this simulation it's telling us  that we can arrive up to 20 kilohertz let's see in   practice digital simulation okay in practice you  will arrive up to 20 kilohertz however you make   it much more robust you know we are right 10 or  something like that and that's why you also reduce   the dynamic range that you need a higher frequency  understand why this is a this is a very important   so the more the speakers we use the more robust  we make the system also too inaccuracies on the   camera system because the camera system we try we  trust it but sometimes it's not accurate you know   yeah sure sure but it's very and this is basically  just to show you how this emulations that we were   putting before how they will work with with an  array with an array you have many more speakers   so whatever you're doing becomes much more  robust yeah so once you have an array this is   what happens okay so this is the comparison the  blue radiation pattern is the radiation pattern   that acrossal cancellation system obtains one  tries to create virtual Health okay and you can   see it's going to make a beam which is very very  strong in one year and a null a notch yeah yeah   okay um in theory you see this is this is how how  something will look on the free field at a given   frequency in theory the array having two speakers  or nothing in the red there is red radiation   pattern has 16 speakers they should work the same  to create this vertical headphones effect you know   however we don't listen to some Nanook Chambers  okay we listen to things in rooms that have   Reflections and that are affecting so by using  many more speakers we create a narrower beam and   this is what is the beam forming effect and this  is why when we have more speaker we focus on and   we create a bubble of sound you know and that's  why the example the the Razer somebody says it has   3D audio game for me because it's a combination  between crosstalk cancellation and the information   nicely very much and just uh Simon to show you  basically this is what happens in real time we   have these beams okay and this basically we have  I mean for the left ear and I mean from the right   ear these are the the ears of a user here these  two loads if the user moves the position of these   beams are changing in real time right right okay  because we are changing in real time the filters   that are creating those beams in the speaker  array which is what this thing here yeah very much   fascinating okay yeah man I think this  is the end of my presentation yeah and that's our vision we want to be the Navy and  technology behind personalized and immersion for   your reproduction offering at least in a  revolutionary listening experience in a   multitude of everyday situations without the  need of a headphone you know yeah I mean at   CES I got the chance to hear the demo with  the tectonic audio Labs drivers and I mean   I could say that you I mean I don't want to  do it right here because my I'll go off mic   but basically you know it really does kind  of track your head and follow you around and   left and turning right it captures all that  kind of experience it's it's quite impressive we use their the tectonic technology because  and like this tectonic BMR drivers are a bit   more point source like it improves at higher  frequencies and how our team already work because   basically they yeah they work a bit better you  know yes and then yeah that's that's that was   a good demo and yeah and you can't really  feel it so I don't know if Simon you have   the opportunity to listen to our technology  one day yeah when it sounds like um I mean I   can tell you but once you listen to the system  you say wow you know I'm looking forward to it   all right yeah okay so uh I guess uh if people  want to get in touch for uh technology they can   uh I'll put your contact information for the  right person or yourself below um yeah you can   contact me you can contact anybody that's that's  good yeah okay all right Simon did you have any   follow-up final questions I'm good I'm good all  right okay well we encourage everybody put your   comments down below questions and anything else  that you're intrigued by please uh please uh   give us a like subscribe and share and all that  good uh YouTube and now Spotify and all the other   platforms so uh Marcos thanks for joining us today  we appreciate your time yeah I think thank both   of you for hosting me and it's a pleasure and I  look forward to seeing the video yeah thank you   very much it's cool all right yeah all right okay  thanks everybody we'll see you next time bye bye

2023-02-13

Show video