NETINT Technologies about India & the rest of the world – difference in technology distribution

Show video

Welcome to NETINT's Voices and video where  we explore critical streaming related topics   with experts in the field. If you're  watching and have questions, please post   them as a comment on whichever platform you're  watching. We'll answer live if time permits,   otherwise we'll respond after the show. Today we chat with Krishna Rao Vijayanagar,   who is founder of OTTVerse, which covers  OTT technologies and business topics. By   way of education Krishna has an MS and PhD in  electrical and computer engineering from the   Illinois Institute of Technology. I met Krishna  years ago at one of the several positions he's   held in both US and Indian companies. Today our  conversation will focus on streaming in India,  

which now has the largest population in the  world. We'll cover the market forces that impact   OTT strategies in India and then focus on the  technical decisions that these market forces   entail, like codecs selection, ladder formation,  player support, CDN selection, and other issues.   Then we'll spend some time discussing  OTTVerse and particularly the issues that   OTTVerse readers are finding most compelling  today. Krishna, thanks for joining us.  Thank you so much, Jan. Thank you Anita  as well for having me on this podcast.  Okay, well we're glad to have you and I know  it's late for you and we appreciate your working   with our time constraints. But you and I spoke  about this a couple days ago. It seems like the   Indian market is a very challenging market  to get into in a number of ways. Could you  

cover that? I mean, what's the Indian market  like to get into as a content provider?  India is a very beautiful, complex and ancient  multicultural country. That creates an atmosphere   of challenges and opportunities as well. To give  you an example, we have approximately 26 different   languages each with their own script, each  with their own vocabulary, each with their own   dictionaries. So when you want to communicate with  somebody, and all the states in India are divided   based on linguistic, based on the languages. So  I come from Bangalore, the tech capital of India.   If I go to my neighboring state, there's a very  small likelihood I'd understand their language.   So if you think about this in terms of OTT, if I  produce a movie in my language, distributing it   to other parts of the country becomes a challenge,  primarily because people wouldn't understand it. 

So this, though it looks like a challenge, it's  very interesting from a technical perspective,   brings in interesting concepts of dubbing. Can  you create one language and translate it or dub   it in five or six different languages? Can you  do close captioning? Can you do subtitles in   different languages and different scripts? How  do you deliver all of this? I was just thinking   about this with a couple of friends after our  discussion, and we realized that several OTTs   have to struggle with the UI. So in the US you're  probably accustomed to having the UI in English.   In India, most of the OTT players have  a English UI, but how do you change the   entire UI to a different language, and how  do you do it for five different languages   in India if you want to be a national player? So it's a beautiful complex country, a lot of   challenges, and it's a good place to be if you  want to solve interesting problems at scale.  

So I leave it at that and we can pick  up these topics one by one, I guess.  Okay. What about pricing? Because it felt like  pricing was also pretty pressured over there.  I would say most vendors who have probably come  to India have also faced this. So India is a   price sensitive market. It's primarily,  because the population is very young,  

they're coming out of the towns, villages coming  to the metros. Recent stats show that 65% of the   population is below the age of 35, so it's a very  young population earning their first dollars,   getting their first paychecks. So people want  to pay less for content. People appreciate free   stuff in India, so it makes it quite complex  to run a business over here. How do you produce   content? How do you set up the technology,  how do you deliver it? Do the marketing,   while charging something which is sensible. So for example, Disney plus Hot Star,  

I think their top tier comes  roughly to a few dollars per year,   probably 10, $20 a year. So that's how competitive  it is. On the other hand, you have forces like Geo   Cinema who's running this year's IPL and running  their entire app for free on a pure reward model.   So this is a conglomerate with very, very deep  pockets. So they can do a market exercise like   this to capture the audience. So if you are an  OTT, if you want to enter the market, how do   you actually counteract such a force? So that's  something interesting to think about as well.  Who's doing a good job coming in from the  outside? Here in the States we have the big names,   Netflix Prime, Hulu, Paramount, who's doing  a good job penetrating the Indian market and   what are they doing to accomplish that? So when you look at the Indian context,   it is interesting. You have to look at it as  the national play versus the regional play,  

just because of the language differences that  I spoke about. So if you look at the national   players, there are the likes of Zee, SonyLIV,  you have ALT Balaji, you have Amazon Prime video,   which does a very good job aggregating content,  JioCinema, so these are interesting players.   Then you have the regional players like  SunNXT, which is for the Tamil market, AHA,   which is Telugu, you have Marathi, you have Planet  Marathi, then you have Gujarati, ShemarooMe. So   you have a bunch of regional players who are  doing fantastic work capturing their audiences.  And then you have a very interesting phenomenon in  India where the telcos are aggregating all these   OPTs. So you have these super aggregators, so  JioCinema is one, Airtel Xstream is another one,  

Tata Play Binge, Hindustan Times, OTT Play. What  they do is they tie up with all these OTTs, they   provide all the content on a single app,  through a single searchable interface.   So they charge a single price. I remember Airtel  Xstream would charge roughly one and a half  

dollars a month to access 15 OTTs, more than 3-,  4,000 titles. So it's a price war on one side,   market forces. How do you compensate the content  creators? How do you still play your bills and   keep the lights on? It's a market trying to figure  itself out and it's in a very interesting phase. 

Let's start to look at some of the  technical issues that producers   face. What's the typical bandwidth that  people are streaming their videos over?  Again, if you look at it geographically, we  typically divide India into the metro cities,   which are the largest ones like  Delhi, Mumbai, Calcutta, Bangalore,   Chennai. These regions typically have pretty  high bandwidth, good mobile penetration,   all of this. Then you have the Tier one, Tier  two, Tier three cities where the bandwidth drops.   But from a business perspective, many people  would agree that the metros are saturated.  

People have two, three subscriptions and  that's not where they're going to get new   subscriptions. They have to penetrate into  the cities and the towns which do not have   very good connectivity. So this goes into the  mind when you actually decide bitrate ladder,   if that's what you are alluding to,  the bitrate ladder and the bandwidths. 

So typically you would see people streaming at  720p around the 1-1.25 Mbps range and a lot of   lower bitrates like the 640p, 480p, 360 goes down.  Your 1080s are generally between 2 and 3 Mbps and   with the caveat that I'm only talking about H.264  right now. For mobile consumption being around 2.5   about until around 3 Mbps. And then when it goes  to large screen, it comes to our 5, 6, 7 Mbps. 

Looking at the encoding ladder, are companies  doing a different encoding ladder for metro   areas and also for the outlying regions,  or do they have one encoding ladder   that's going to serve everywhere in India? From what I know, and there might be a few   exceptions to this, it is typically on the device  profile, rather than the geography. This is what   I've seen. So you have these per device manifest  sort of configurations. Where when you're using   H.264, you cap it at close to 3 Mbps for 1080p,  and you can go higher for large screen. And I  

should add that India is a mobile first country.  There was a report by Hotstar two three years ago,   Disney + Hotstar where they analyzed an entire  year streaming and they said that 94% of the   subscribers accessed it through an android form,  through a phone, and most of it was Android. iOS   has a very low penetration just because of the  price point, but if you look at it that way,   it doesn't really make sense going up, having  very high bitrates, HDR, all these capabilities,   when 90% of your subscriber base is  going to access it on a handheld phone.  Typical ladder might go up, what's the top bitrate  you would see? I think you said five megabits per   second. Is that where you're capping out? Yeah, five, especially if you're on the  

large screen. But with the OTTs I've worked  with and I've inspected just out of curiosity   over the last few years, it's typically  been 2 to 3 Mbps the top bitrate. The   quality on your mobile phones is fantastic  on the laptop, it's great. It does the job.  What's the player situation look like? You  mentioned that most of the playback is on   Android phones. Apple is very premium,  very costly, here in the States. What   about smart TVs and OTT devices, even in the  big cities, are those making penetration?  They are. So I was in a talk a couple of weeks  ago where a person from Samsung Ads actually   said that there are 14, 15 million TVs being sold  every year. So it's not well penetrated in the  

Indian market, primarily because it requires data,  you have to end up using your bandwidth or your   internet connection. So DTH and normal linear TV,  are very, very prevalent in the towns and villages   of India. Smart TVs are catching up as data is  getting cheaper and as bandwidths are improving.   So we aren't there yet. Cord-cutting isn't isn't  such a big phenomenon as it is probably in the US.  We've talked about the streaming side,  what about the broadcast? I mean,   is there a broadcast infrastructure over the  air that makes up a substantial percentage   of what people are actually watching? So Cable TV, IBTV is very, very common   over here. It isn't as expensive. I remember when  I was in the US I had a Comcast Xfinity connection   a hundred dollars a month and I'd get internet,  probably a phone and a lot of channels. It's  

really not that expensive in India, it's probably  300 of 400 Rupees, which is roughly around $5. And   you get a host of channels, probably a 100, 150  channels. And again, it was government regulation   that you actually get a sheet of paper to your  house where you can tick and you can mark what   channels you want and only those will be provided  to you. So I still remember doing that. So in our   house we speak couple of languages, just because  how my parents came together. So we just choose   channels from those two languages and we're  done. We don't need all 100, 150 channels. 

And that's all MPEG2 or H.264? I believe it's MPEG2.  What about Codecs over there? I mean I heard talk  a few years ago about VP9 making good penetration   because A, it's supported on Android devices  and B, it's more efficient than H.264. Are you   seeing other codecs being used over there? I was speaking to a friend a few hours back   and he said yes, there's a difference between  penetration and adoption. People are trying   to adopt new codecs like HEVC, but generally when  you deploy, when you're deploying on smartphones,   AVC is good enough, you don't have any royalty  issues and for the bitrates we're streaming, I   think it does a perfectly good job. Plus hardware  support is there, it's a legacy codec by now.   So a large number of these mobile devices aren't  your high-end Android devices. It's not your   Samsung or your OnePlus devices. A lot of them  are Chinese phones, which you get for roughly $50.  

So massive consumption on those devices as well.  So anyone trying to deploy a codec has to keep   this in mind. They might not have the best DRM  support on board as well, so that restricts you   on what you can actually do. You might not  be able to serve 1080p on certain devices,  

just because of the hardware capabilities. We talked about the high end of the encoding   ladder, what's the low end of the encoding  ladder, what's the resolution and bitrate?  I have seen it go down lower than  360p at some cases, but 360 is   roughly where I see it cap off, 360 at roughly  200, 300 kbps. I don't see what people could   make out of that resolution, but yeah, that's  lower end to keep the streaming going I guess.  Yeah, I remember when 640 x 360 was pretty high  quality. When you're streaming at 200 or 300 kbps,  

it feels like you want a pretty  consistent stream. Are you seeing   a lot of CBR over there or is it 2 pass VBR  or even exotic things like constrained CRF?  So speaking from experience and having been  a vendor in the compression space, people do   pick up, I would say platforms like Elemental, you  have few people running theirs off of Brightcove,   certain other vendors in the space. So whatever  they've provided is typically what they use,   but people who do-it-yourself or choose certain  vendors in India, I've seen that the common use   case is either CBR, not 2 Pass, but plain CBR or  capped CRF. I personally have dabbled with capped  

CRF in my previous role, we found it to be a  very good approximate to CBR in terms of quality,   it was very good. It didn't overshoot the  constraints that we had put too badly, and   was perfectly good enough for streaming in India,  without any buffering issues or any of that stuff.  Can you just briefly describe what CRF is and  what capped CRF is for people who may not know?  CRF, so we are talking  about constrain rate factor.  Constant rate factor. Constant rate factor, right? So this is a mode   in FFmpeg, which is probably the most popular open  source implementation of video codes out there,   where essentially they try and hit a video  quality by adjusting the number of bits allocated   throughout the video sequence. This is in contrast  to CBR where CBR tries to maintain certain bitrate  

by adjusting the video quality. So you would  start off, let's say you start off with an I-frame   that would get the highest quality, then your  P-frames would get the most allocation of bits   and then it'd be distributed amongst the B-frames  trying to hit your bitrate caps. CRF obviously   has to do similar stuff, but it places a greater  emphasis on the quality, rather than maintaining   a very constant bit rate. So you might see it  overshoot, you will see it overshoot actually.  So modification of this is the capped  CRF where you can put a cap and tell CRF,   maintain your video quality while not exceeding  a particular bitrate by a certain percentage.   So this makes it very suitable for ABR streaming.  And if somebody asks why to that question,   I would say bitrate is in my understanding, is  a contract between the player and the server.  

So when the players is told that this stream is  5 Mbps, it expects and believes that the server   will send five megabits per second. Anything more  or less by a large amount will cause problems at   the player. So it's almost like a contract. CRF  will end up violating that contract capped CRF   not much, and it provides very good quality. So for CBR you would set a bitrate and then   you would set the maximum bitrate and  for CBR, those two would be the same.   What do you do in a CRF situation? What's the  setting that you use in the command string?  I believe it's the max that you set the -bv  and then you also mentioned the CRF setting,   so it typically has the CRF setting with zero is  almost lossless pristine quality, which no human   can make out and 51 is really poor. Typically  in practice, we have gone between 20 and 25  

and this is through just golden eye testing and  VMAF scores where we realize certain genres don't   require 18,17. We didn't see any difference on  a mobile phone. So this is why I go back to the   business side of things, where you understand the  geography you're streaming at, the devices you're   streaming to, and if you realize that 95% is going  to be consuming on an Android phone, there just   isn't a lot they will make out if you change your  CRF setting from 24 to 25. You might save a lot on   your CDN card bills and your storage bills, but  you're not harming the user experience at all.  Yeah, you and I talked about this issue just, I  guess, a couple weeks ago I gave a presentation   at Mile High Video earlier this week and I sent  you some 2 Pass VBR end codes and you said, well,   that it looks great, but you're doubling  your encoding cost, and then you relayed   your experience where you used capped CRF, with a  CRF value of 25, I believe, and then the maximum   setting. Thank you for that, because it was a  really interesting portion of the presentation,   but we did the overall VMF  scores were about the same.   The encoding time was cut in half. There were  some variability issues. I sent you that data,  

but overall it seemed like a pretty good strategy. We talked about VMAF, or you mentioned that really   briefly, are you using the phone version or the  phone profile for VMAF? Have you experimented   with that or are you using the default? So again, we do this sort of comparisons   where we run it on the phone model and then on  the desktop model, I know there is a 4K model,   or am I wrong on that? There is.  One, right? I've never tried that, to be  honest. Was never in the situation of test   4K. So the phone model and this, and we actually  see a very big difference, you could change your  

bitrates quite a bit and the phone model, the  scores wouldn't budge. So we realized that hey,   here is something that could actually save  us a lot on CDN bills and storage costs.   And that naturally leads to per  device manifests doing something,   figuring out who's asking for a manifest and serve  the right manifest, doing edge manipulation of the   manifest. So I think it naturally tends itself  to a lot of innovation, just having those scores.  And how do you see that working? I mean, what  different profiles are the typical publisher   supporting? You've got a mobile profile,  is there a smart TV profile at this point,   or it is computer playback? I guess, it's  not such a big deal for premium content   here in the States, but what about in India? I honestly have worked to only a couple of   them who have seriously thought about per device  manifests. It has been tried, but if I look at...   The premium publishers, yes,  they are looking into this,   they are doing experiments, they are delivering  it this way, but then you have a bunch of   OTTs which are very sure about where  their audience is, and they know that   quality is good at certain bitrates, so some  of them don't have these considerations also. 

So HDR you said, there's not a lot, I mean, unless  you're distributing to the living room and unless   you're using HEVC, HDR's not a factor. What  about DRM? Is that typically used over there?  DRM is a big thing. So either people roll out  their own using this AES-128 or they go with   Multi DRM vendors who provide them with  the common Widevine, Fairplay streaming,   and Microsoft's Play Ready. There isn't, I would  say, major innovation going on in the DRM space,   it's just pick up a vendor and go with it. And what about the CDN side,   are there multiple CDNs that are covering the  different regions, or is there one big CDM that   everybody uses? How does that work? So Akamai and CloudFront are   pretty popular in India. I've seen several  companies do multi CDN switching between   these two. You also have AirTEL who's entered  the race with a tie-up with Qwilt to role out  

their own CDN and Jio's also doing the same.  So you have these local, the homegrown vendors,   and plus you have Akamai and CloudFront.  These are the big guys in the market.  Are people doing multiple CDN  support with switching and-  It is becoming a thing over here, but to be  very frank, and I have been looking at data   also for the past few years, CDN's have become  very reliable and very good at streaming. You   only hear of an occasional hiccup and  crash. They are pretty good right now.  

You might do multi CDN for certain issues probably  in a Tier two, Tier three city where certain   transmission isn't very good or to save costs,  I might have a better rate from A versus B and   then I switch them during the prime time. It sounds like overall the market is a   challenging market to get into, because  of the languages and the regional,   but the encoding picture is actually pretty  simple. I guess it's not... It's relatively...   It's not that hard. Multiple codex are not a  thing. HDR isn't a thing, exotic encoding ladders   aren't a thing. So it seems straightforward,  the content play and getting people to buy   your stuff, I guess, is the big thing. Yeah, so it's the business side. Sorry   to interrupt. It's actually the business side  which is being on all these technical decisions.  

So I'm sure a lot of the teams are dying to  deploy a HEVC VP9 and try the data stuff,   but does it actually pay off? That's the question. What percentage of revenue, if I'm an Indian   OTT provider, what percentage of my  revenue and percentage of revenue,   not number of subscribers, what percentage  comes from inside India and outside India?  Oh, John, probably I don't have the  right answer to that question, but-  We were talking and you were mentioning  that there's a huge market outside of   India for all the people who have left it, and  they're paying US prices, they're not paying-  They are. Yeah, so that point is true. I don't  know the exact split, but definitely Indians   are there across the world. I think the biggest  hotspots are the US, UK, Canada, Middle East,  

and a sizable population in Africa as well, South  Africa. So people streaming to these regions,   they end up paying in the local currencies. I  think the example that I was talking to you the   other day was a particular streaming service which  charges like $60 Canadian over there and probably   $2 or $3 US in India. So you can make a lot of  money out there, and it's also profitable to   stream outside India, just because of the add  CPMs. You can earn several dollars, in double   digits probably in the US, whereas in India it's  probably $1, $2 CPMs. That's where it caps off at. 

And just, I should have asked this back  when we started talking about the content,   but the subscription rates are low. Is it  advertising supported, or is it subscription only?  Most of OTTs today do a hybrid, or are  going towards hybrid having started off   only subscription. So in India it's a thing to  have a mobile only plan, where you can play back   only on a mobile device. You can't play back  on anything else. So that's your lowest price,   which might come with ads, might not come with  ads. And then you say, okay, here's another tier  

where it's on large screen, it's on mobile,  but you will see the occasional pre-roll ad,   mid-roll ad and if you want to turn it off, go to  the premium plan, add free streaming for one year.   That's typically what happens over here. I mean,  the IPL is an oddity in this entire screen where   they're streaming everything for free with an  add supported play, but there have been talks   that they'll be rolling out a subscription plan  in the next couple of months, after the IPL. 

Let's switch gears and talk about OTTVerse. I  had my site streaming earning center and I've   contributed a lot to streaming media over  the years. You kind of came out of nowhere   and really grabbed a big share and in a big  way. What was the idea behind OTTVerse and   what has it grown into over the years? So the story is very simple. I was tired   of working 12 years on the trot. So when  the pandemic hit, I just took a break and  

I started writing. I had always blogged  and written about obscure stuff, recipes,   cricket and stuff like that. So I had a lot  of notes lying around. So I cleaned them up,   put them online, and it suddenly became popular.  People approached me to sponsor articles and put   ads on the website and then one thing led to the  other end, we decided to make it a business. So   that's how it started off. I think our foundation  and our goal is always spreading knowledge.  

I found it very tough to switch, I was in Harmonic  career, so I found it very tough to switch between   transporting and OTT. I just couldn't find proper  material which would go in depth. So I tried to   just put my notes online and help others as well. So what statistics do you share about viewership?   How many eyeballs? Geographically we are 55% US and EU put together,   20% India and the rest is spread across  Asia and Latham and the UK. That's our   geographical spread. It's mostly from the US and  the European Union. Then in terms of viewership,   we are flirting with a hundred thousand views  per month number. Sometimes it's up, down,   we'll probably be consistently above  that in the next couple of months. 

What's the distribution? I mean who's looking?  Is it big company, small company, developer,   OTT programmer? I mean, who do  you see coming to your site?  That's interesting actually. So a lot of  statistics actually we get out of LinkedIn,   because we have a pretty active channel over  there. So what we've seen is when it's deeply   technical, we have the folks who love tech  who come in. So it might be a programmer,   it might be an entry level engineer, it could be  as high as a CTO, but mostly on the tech side of   things. And we have a lot of explainer articles  which kind of simplify stuff, like what does   client side add insertion. So I've actually met a  lot of marketing teams who read that and find it   easy to understand and then explain it to others.  And then we have also started a few opinion pieces  

interviewing OTTs business owners. So that  obviously attracts a slightly different crowd,   the C-levels and directors. So we have a healthy  mix and I think that's how we'd like it to be.  What other, you and I talked about some of the  educational initiatives that you're producing   from OTTVerse, what are you doing there? Inspired by a lot of your workshops, I think,   truly there is a need for concentrated workshops  on probably using FFmpeg for compression for also   pre-processing, post-processing in terms of  packaging, actually putting DRM together.   A lot of this knowledge is hidden and it doesn't  have to be, because FFmpeg is also open source.   Why can the knowledge be also open out there? So  these are where we want to do a bit of educational   initiatives, and specifically targeting colleges  in India, universities, because video isn't   really a big topic which is taught over here.  So they go on the more mathematical concepts,  

like image processing, signal processing, they  skim over the video, the entire topic of video,   despite it being a super complex place, you can  probably enter and exit, you can retire working   on this OTT pipeline. There's enough work for  the next 20, 30 years. Students aren't really   aware of that when I interact with them. So  that's something that we want to do as well.  So what does this translate to? You and I  talked about potentially doing workshops in-  Correct, so this is probably going to be a  couple of days workshops, if it's in person,   go down to an office. Oh sorry, that's the other  thing, right? We see a lot of engineers coming   into the workforce not really understanding what  is OTT. Specifically, I would even go around   saying that teams when they test, the tests are  pretty simple. I press play, if it doesn't play,   hey, it has a bug. But can you go a little  deeper? Is there a problem in the manifest,  

or can you create test cases by deleting certain  files, by having them on the manifest. So when   the player tries to play it, something happens.  Is the player supposed to crash? Is it supposed   to fail [inaudible 00:32:14]? We don't know,  right? So just the general education, either   in the corporate region or the educational  region. It could be a couple of hours,   or a half a day online course, or they actually  go in person and walk them through an entire week,   talking to them about end-to-end, right from  content upload to playback, recommendation,   search, and then get their hands dirty. It's not very difficult if you have like   five hours on your hand, you could learn  compression, how to use FFmpeg command line,   do the HLS, set up a server and actually stream  it to video.js. And I think when somebody  

presses play and it actually works, that much is  enough to encourage them to take the next step.  Does light you up when it actually works.  What are the topics that your readers are   finding interesting today? Looking at your  page views, what topics are most compelling?  We find a healthy mix between transcoding and in  transcoding we see lot of questions still on CBR,   CRF, 2 pass. There are very good articles  online. I mean I don't say mine is the best.   So there are very good explainers. There are  good guides from the FFmpeg website itself,  

but people still want to understand more. They  ask about the fundamentals of compression,   like what's an IDR frame? What's a CRA? What's  an IPB? People want to understand that. Those   articles are pretty popular. How to package that  seems to be trending. Then it goes on to DRM and   ad insertion, client side, also server side ad  insertion, because these are the jargon which   are thrown out there in meetings. So people  want to understand what's the difference,  

how do they work, and stuff like that. Getting some questions in. One question,   I guess, more specifics on the publishers who  are succeeding in India, wanted to know how   is Netflix doing, how is Prime doing, how is Hulu  doing? Do you have any information to share about   how successful those companies have been? I know  you mentioned Prime seem to be doing pretty well.  Prime is doing well. They have a lot of original  series that they've produced and released.  

So Prime's good. Netflix hasn't had that sort  of penetration, because they're primarily,   I would say, a premium platform, price point is  also high. They were forced to reduce their price,   because they didn't find any traction in  India. So they have a mobile only plan,   which is 149 Rupees. Prime is still restricted, I  would say, from my guess is to the metro and the   English speaking pockets of the country, but it  won't have many takers in the large part of India   where 80% of the population lives. But Prime  is big. Jio has recently tied up with HBO  

and Warner Brothers, so all of that can now be  accessed on Jio in a couple of months, I suppose.   They were early on Disney Hotstar, but that  relationship broke and they moved over to Jio.   Hollywood content is also available  through Lionsgate, Lionsgate Play.   I don't know if they have an app of their  own, but they are available on aggregators.   They are available on Airtel Xstream. A question came through about what's   being used in terms of packaging. Is it more  HLS or Dash? I guess it's more Dash, right? 

It's actually both. Funnily enough, most of the  news channels in India also livestream onto their   websites and all of these are unprotected  HLS. So HLS is big and Dash is also big.  What about CMF, is that making  any penetration at all? Or just  Because of the fact that a single DRM  cannot be used, I don't see lot of,   a single file format across all DRM. I've spoken  to somebody a month ago and they said that it's  

still not on the radar to try a single file. Just kind of an off the wall question,   what has been the impact of the success of Slumdog  Millionaire? I guess that's a few years back, but   I mean that... How did that change? Is  it the perception of content in India, or   what changes did that deliver? I don't know from the Indian context, it was   just great that an Indian movie won an Oscar. But  a lot of things, for example, the music producer,   A.R. Rahman who won I think the best music as well  for Slumdog, he composed 11 songs. This is a guy   who from 1992 has probably been producing like  50 to 100 songs every year, every song different,   across different languages. So when Rahman won,  everybody was, high time he won something. So I   would say it had a massive impact, because  the Bollywood industry has been a thriving   industry for 70 odd years. Several superstars  who have been popular across the world as well. 

How do they do it? I How do you produce a movie,  Slumdog Millionaire was one language in the   US. How do you do it for dozens of languages in  India? What does that look like in the theater?  Honestly, it's a voiceover. For most movies,  it's where you have multiple dubbing artists.   So I've seen two, the most common phenomenon is  you shoot a movie in one language and you dub. So  

you have dubbing artists for every actor. So it's  the right language, the right accent, all of that.   Or the movie is completely re-shot. I've seen  that as well. So you have a very popular series   of a super cop, Dabangg. In one language the  storyline is the same, but it's one actor who's   popular in that region and then it gets shot  in another language, same storyline, but with   a different actor. So it's either that or very  expensive ways to actually have the same cast,  

the same background, everything. Have different  people coming shoot their parts and move on.  Organizational nightmare. We have a question  about NAB. You were at the show. What did   you see that was kind of impressive to you? Before NAB I was talking to a few friends and we   were kind of guessing what is the next big thing  from the India context at least. We realized that  

OTT content, the same content is going to be on  multiple platforms very soon. It's either Sony   on Jio, or SonyLIV on Airtel Xstream on Tata, now  it's the same content, same time of the day. You   get it on the same day. What makes you decide  where to go? Is it price or is it content? Is   it quality of the app, or is it the quality of  recommendations? So that's when we realized that   probably the next thing which people are trying  to crack for many years is churn reduction,   personalization, and reducing the amount of  time you have to spend searching for something.  

I think that's a persistent topic in every  trade show. How do you reduce the amount   of time? I spent nine minutes trying to  find a movie. How do you reduce that?  So that kind of stood out this year at NAB, at  least for me. I saw a couple of companies Think   Analytics personalizing the EPG page, which  to me struck a call. Because in my house we   subscribe to probably 20 channels on our cable,  but we still have to scroll through like 600,   700 channels or memorize all the numbers. Now,  why can't you just personalize the EPG for me?  

I watch those 10, why can't you just pull them  up to the top, or create my own channel? Virtual   channels are now becoming common, what to live. Have you covered the personalization side?   Because that's not something I've looked into.  I know it's a thing, but I've not ever written   about it. Pretty hard to test, I guess. A couple of articles. We actually have a  

contributed article which will come out next week.  I won't mention who, but it'll be coming out in   the next two weeks. It's by a popular OVB. They're  talking about personalization and data collection.   Where's the boundary, how much data do you collect  to personalize. So it's a very interesting take.  One last question. Did you see anything AI  related in the codec space that you thought   would be impactful in the next two to five years? So I've always loved the use of AI in codecs   and it doesn't have to be super complex. I've  done tests myself where you can actually use  

a genre of codecs and you'll know what are the  popular settings which work very well for them.   That's itself a use of machine learning. Thousands  of parameters which are generated every frame,   if you can run them through AI is a buzzword  to me. So I typically say machine learning,   which is the more technical and the right  way, but if you run certain algorithms on it,   it makes your next compression easier. So that's  where I would love to see everyone go towards.  Scene understanding, understanding the different  scenes of a movie, which I believe couple of them   are working towards. A movie can have multiple  scenes where you have people sitting in a coffee   shop. Can it dip into a repository of settings  to come and compress that particular scene,  

versus then somebody's running behind a football  and use a different set of settings over there.   I would love for some machine to be able  to look at every fame and say, this is the   ideal bitrate distribution. Go for it. Do it. Interestingly, at Mile High there were a bunch   of codec vendors, well not really even codec  vendors as much as encoding, and Netflix talked   about what they were doing. Sky talked about  what they were doing and it feels like there's   going to be a lot of machine learning based  innovations coming out in the next two to five. 

I don't know how it's going to hit something  like FFmpeg. It's going to be interesting to see,   I mean, you and I specialize in making technology  usable to the average Joe, I don't know, Netflix   showed some AI based scaling technologies that  it's great, but it's not something that you can   access from Ffmpeg, unless they open source it. I would actually suggest a different route,   perhaps something like the VMAF library, where  you don't have to understand what happens inside,   because it's pretty complex, but you can  actually use it. So if somebody converts a   database or let's say a hundred thousand clips,  and then you are able to somehow connect FFmpeg   to that database, that's good enough. It  understands, sends some statistics there,   that server sends back saying, here are your  optimal bitrates, go for it. That itself is good   enough for 99% of the population, I would say. I'll keep an eye hour for that. Listen, we're  

out of time and we're out of questions. I really  appreciate you spending time with us. I know it's   late at night, but it's always great to chat with  you and I appreciate your taking the time today.  Oh, thank you so much, Jan.  This has been a lot of fun.

2023-05-20

Show video