0:00:00.005,0:00:05.525 Module 10.3, examples of start-ups that [br]use signal processing as a core 0:00:05.525,0:00:09.502 technology. [br]Earlier on in this class somebody asked 0:00:09.502,0:00:13.012 in the forum, if I follow digital signal [br]processing class, can I get a job in the 0:00:13.012,0:00:18.276 start-up and what sort of start-ups? [br]So this brought us to think, well maybe 0:00:18.276,0:00:23.550 we can describe a few start-ups that came [br]out of research done in the lab. 0:00:23.550,0:00:28.230 And our four that we discuss here, there [br]are actually more that are active, but 0:00:28.230,0:00:32.406 four will be discussed here are [br]Illusonic, Quividi, Sensorscope and 0:00:32.406,0:00:36.615 Vidinoti. [br]So the first start up I want to discuss 0:00:36.615,0:00:40.435 is called Illusonic. [br]It was started by Cristof Faller who did 0:00:40.435,0:00:43.815 his PhD thesis on a time as a [br][INAUDIBLE], and was interested in 0:00:43.815,0:00:51.060 acoustical signal processing. [br]And in particular in multi channel audio. 0:00:51.060,0:00:54.024 So if you do acoustical signal [br]processing, there are questions like 0:00:54.024,0:00:57.670 beamforming, echo control, we just [br]discussed this earlier. 0:00:57.670,0:00:59.458 We used a project of can you hear, the [br]shape of a room. 0:00:59.458,0:01:03.658 When you want to do spatial audio [br]processing, you want to generate audio 0:01:03.658,0:01:08.942 for many channels. [br]Either for headphones or for multichannel 0:01:08.942,0:01:14.024 loudspeaker systems, you may want to do [br]upmix or you take a stereo signal and you 0:01:14.024,0:01:20.440 would like to render it as a 5:1 signal [br]or as a 17:1 signal. 0:01:20.440,0:01:25.840 And, there are tools, of course, where, [br]you can use signal processing techniques, 0:01:25.840,0:01:33.070 for example, to de-noise Music or [br]de-reverb, recording of, person singing. 0:01:34.110,0:01:38.335 So the tools that are used at Illusonic [br]are classic digital signal processing 0:01:38.335,0:01:42.820 tools, plus what was discussed briefly, [br]when I talk about the class on audio and 0:01:42.820,0:01:49.142 acoustic signal processing. [br]Again, perceptual models are extremely 0:01:49.142,0:01:52.854 important because the human auditory [br]system is a very sophisticated signal 0:01:52.854,0:01:56.856 processing device, and if you try to fool [br]that device you better need to understand 0:01:56.856,0:02:04.324 how it works. [br]Here is an example of cool application, 0:02:04.324,0:02:09.380 so let's say you have your home cinema [br]and you have a stereo recording that you 0:02:09.380,0:02:16.410 would like to listen to. [br]So the home cinema has actually in this 0:02:16.410,0:02:21.435 case one, two, three, four, five, six, [br]seven, eight, nine plus probably two base 0:02:21.435,0:02:26.310 booster somewhere, so it's probably an [br]eleven channel system, so you would do 0:02:26.310,0:02:34.460 enough mix from a stereo signal, let's [br]say from your MP3 player... 0:02:34.460,0:02:38.051 To this eleven channel spatial audio [br]system, and you would like to make it so 0:02:38.051,0:02:42.000 that it sounds really like you're in the [br]concert hall. 0:02:42.000,0:02:44.912 And so even sony has a very cool [br]technology to do this, and not only do 0:02:44.912,0:02:47.928 they have the technology, they actually [br]sell a box that will do this at 0:02:47.928,0:02:55.182 professional quality level. [br]So the company is it's a small company 0:02:55.182,0:02:59.403 about five people, half a dozen people, [br]it licenses technology, state of the art 0:02:59.403,0:03:05.775 stuff, to other, companies, and it has [br]custom technologies that it develops. 0:03:05.775,0:03:09.935 For specific applications and as I [br]mentioned it has this very cool Immersive 0:03:09.935,0:03:14.160 Audio Processor that was just launched [br]this year and please visit our website 0:03:14.160,0:03:18.060 and see this cool stuff and if you want [br]to buy one of these Immersive Audio 0:03:18.060,0:03:24.660 Processors, I can tell you it sounds [br]incredibly beautiful. 0:03:27.010,0:03:29.050 The next company I want to describe is [br]Quividi. 0:03:29.050,0:03:32.634 Now this is a very important company in [br]its class because its a company of Palo 0:03:32.634,0:03:36.794 Prandoni. [br]So when he's not teaching on Coursera and 0:03:36.794,0:03:41.530 playing his his guitar To explain signal [br]processing. 0:03:41.530,0:03:46.173 He's actually the CTO of a company in [br]Paris, called Quividi. 0:03:46.173,0:03:49.935 And Quividi does a full length thing in [br]environments where you have cameras and 0:03:49.935,0:03:54.017 you have digital signage. [br]So we have advertisements on screens or 0:03:54.017,0:03:57.702 you have information on screens, then [br]Quividi clearly allows you to monitor who 0:03:57.702,0:04:01.160 is actually watching what you are [br]showing. 0:04:01.160,0:04:04.688 So if you have a bunch of people in front [br]of this camera, it will identify also 0:04:04.688,0:04:08.216 people it will say oh, here is a lady, [br]here's ladies, you also got, a few of the 0:04:08.216,0:04:12.908 people are guys. [br]It will make some statistics, how long 0:04:12.908,0:04:16.604 the people actually watch for example in [br]advertisement, where they look on the 0:04:16.604,0:04:21.547 screen and so on. [br]And this entire system is distributed in 0:04:21.547,0:04:26.335 the cloud, and Allows you to do a [br]dashboard, a so-called dashboard, of how 0:04:26.335,0:04:31.579 your advertisement is being seen in these [br]public screens, or in the malls where the 0:04:31.579,0:04:39.392 screens are being shown. [br]And at latest, they have 150 networks of 0:04:39.392,0:04:45.334 measurements that are deployed all across [br]the world as you can see. 0:04:45.334,0:04:50.570 And there are some very famous names that [br]show up and so they essentially can do 0:04:50.570,0:04:55.113 monitoring of the quality of [br]advertisement for all of these companies 0:04:55.113,0:04:59.964 essentially in real time and provide [br]reports to the effectiveness of using 0:04:59.964,0:05:08.680 advertising on screens in public spaces. [br]Okay, that's the story of Quividi. 0:05:08.680,0:05:11.582 It's cool technology. [br]It uses computer vision, image 0:05:11.582,0:05:15.222 processing, the, it also uses a lot of, [br]you know, state of the art, algorithmic 0:05:15.222,0:05:20.344 and machine learning technology. [br]Please visit their website if you want to 0:05:20.344,0:05:26.200 know more about this one. [br]The third company is called Sensorscope. 0:05:26.200,0:05:30.808 It grew out of all the efforts of doing [br]environmental monitoring and various 0:05:30.808,0:05:34.929 projects here at DPFL. [br]So, if you want to do environmental 0:05:34.929,0:05:38.346 monitoring, it's cool if you can do real [br]time visualization of what is happening 0:05:38.346,0:05:42.204 there. [br]an application where people are very 0:05:42.204,0:05:46.551 interesting is so-called precision [br]agriculture, so we want to control the 0:05:46.551,0:05:51.230 quality, let's say for example, of water [br]systems. 0:05:51.230,0:05:55.991 you also want to detect you know, certain [br]weather patterns and so on and then 0:05:55.991,0:06:00.729 optimize crop production thanks to this [br]monitoring. 0:06:01.750,0:06:06.175 So the company does large scale sensor [br]networks, deployments and data 0:06:06.175,0:06:10.370 management. [br]So you need wireless sensor networks. 0:06:10.370,0:06:14.470 So these are small stations that talk to [br]each other in an ad hoc fashion. 0:06:14.470,0:06:17.760 So self organize sensor networks. [br]Then you need signal and image 0:06:17.760,0:06:20.314 processing. [br]The usual stuff that you have learned 0:06:20.314,0:06:22.832 here in the class. [br]And of course radio communication 0:06:22.832,0:06:27.595 technology. [br]So here would be a typical example. 0:06:27.595,0:06:32.282 you build a monitoring station. [br]We have seen such monitoring stations at 0:06:32.282,0:06:37.010 class when we have discussed sampling [br]issues with respect to rain monitoring. 0:06:37.010,0:06:42.210 So you take state-of-the-art, [br]off-the-shelf sophisticated monitoring 0:06:42.210,0:06:46.260 communication and so on. [br]You build sensor stations. 0:06:46.260,0:06:49.710 You deploy them in a self-organized [br]network. 0:06:49.710,0:06:52.870 Then from a bay station you talk to the [br]cloud. 0:06:52.870,0:06:56.814 On the cloud, you have all this data, and [br]people that are interested in monitoring 0:06:56.814,0:07:00.410 this sort of deployment get access, [br]privileged access to this data and can 0:07:00.410,0:07:04.006 take statistics and, you know, decide [br]what to do, for example, about their 0:07:04.006,0:07:12.060 precision agriculture project. [br]It's a small company, half a dozen 0:07:12.060,0:07:16.200 people, and probably its main market is [br]precision all, agriculture, even if it 0:07:16.200,0:07:19.920 started, also from a, academic point of [br]view, mostly about environmental 0:07:19.920,0:07:25.298 monitoring. [br]And you can watch their website here, you 0:07:25.298,0:07:29.453 can also watch all the data that is [br]online at climaps.com. 0:07:29.453,0:07:33.297 So all the deployments that have ever [br]been done by the company and by the lab 0:07:33.297,0:07:37.513 are actually available here on, on this [br]website, and you can also use this data 0:07:37.513,0:07:41.605 and you know, do some further signal [br]processing if you're actually interested 0:07:41.605,0:07:49.652 by this topic. [br]The fourth company here is called 0:07:49.652,0:07:53.518 Vidinoti. [br]It's a recent start up from the lab and 0:07:53.518,0:07:56.998 it works in augmented reality, in [br]particle augmented reality on mobile 0:07:56.998,0:08:00.693 devices. [br]So, the core technologies image 0:08:00.693,0:08:05.717 recognition, computer vision. [br]But in a ways that is robust and then to 0:08:05.717,0:08:10.681 also do all these processing in real time [br]on small devices, like mobile phones and 0:08:10.681,0:08:15.061 decide how much processing you do on a [br]mobile phone, how much you do in the 0:08:15.061,0:08:23.188 Cloud or on the server. [br]And Vidinoti has a bunch of state of the 0:08:23.188,0:08:27.270 art. [br]Algorithms on the one side, to do 0:08:27.270,0:08:31.716 tracking and recognition, and also a [br]number of cool ideas on how to do 0:08:31.716,0:08:37.180 augmented reality based on these [br]methodologies. 0:08:40.510,0:08:44.330 So, the technology is essentially [br]cloud-based. 0:08:44.330,0:08:48.478 But there is an iPhone application that [br]you can download, and then you can 0:08:48.478,0:08:52.966 annotate your favorite pictures or [br]newspapers or whatever with augmented 0:08:52.966,0:08:57.590 reality, and this is actually being used [br]in particular in the newspaper industry 0:08:57.590,0:09:01.806 to sort of bring digital content in a, [br]you know, in a funny or attractive way 0:09:01.806,0:09:08.815 onto a medium. [br]Mainly the newspaper that is being 0:09:08.815,0:09:12.280 challenged by, of course by Internet [br]currently. 0:09:12.280,0:09:17.920 It is a small company, less than 10 [br]people currently. 0:09:17.920,0:09:22.130 It has you know, a strong research and [br]development. 0:09:22.130,0:09:26.486 It has also a strong intellectual [br]portfolio based on patterns that has been 0:09:26.486,0:09:30.689 generated over the year around Augmented [br]Reality. 0:09:32.240,0:09:36.022 And if you want to know more, here's a [br]web site, and here is interactive 0:09:36.022,0:09:40.300 application for the iPhone currently, it [br]will be ported to Android within a couple 0:09:40.300,0:09:47.134 of months as well. [br]So, these were example of start-ups that 0:09:47.134,0:09:50.704 used state of the art signal processing, [br]image processing, computer vision, and so 0:09:50.704,0:09:55.842 on. [br]And bring it to the real world, in very 0:09:55.842,0:10:02.178 concrete applications, from audio, to [br]sensor networks, to augmented reality, to 0:10:02.178,0:10:07.660 monitoring of audience in, advertising.