RiverSilica Explains the Science and Art of Tru-Transcode
This Bangalore-based company is gaining attention by offering higher quality transcodes at a lower price than the competition.
Learn more about the companies mentioned in this article in the Sourcebook:
[Note: This sponsored interview was recorded at Streaming Media East 2013.]
RiverSilica was a new face at the recent Streaming Media East conference in New York City, where it told content owners about its high-level transcoding products. The company's founder, president, and CEO, Jagannathan Balaji, sat down for a red carpet interview to explain the technology that forms the heart of RiverSilica's product line.
"We’re a Tru-Transcode company basically, and the art of Tru-Transcode is our mantra," Balaji said. "We’re actually not too concerned about the future standards because the way we have built the product is to make sure that any standards that we move on we should be able to harness the advantage that we have created in the algorithm. So given that scenario, for us, HEVC is around the corner. We’re working on HEVC as well. Given that, I think from a product perspective we can actually put together a secret sauce on top of the new product line as we go by."
While new formats and standards are a concern for RiverSilica, the team is more concerned with giving customers the tools they need when they need them.
"We pretty much would like to be on the ball when it comes to newer standards, the publishing side of things or transcoding HEVC or input, for example, at 4K for example. So, you know, in that perspective, yes, we are working on it," Belaji answered. "We wouldn’t announce a product for the sake of announcing, but we will have the right things that customer wants and the timing has to be right from that perspective; so that’s one of the things that probably I strongly propagate internally as well."
To hear more from Belaji, such as how RiverSilica got its name or why it's able to offer higher quality than its competitors at lower prices, watch the full interview below.
Siglin: We’d like to welcome Jagannathan Balaji to the Streaming Media Producer red carpet interviews here at Streaming Media East 2013. Balaji, tell us a little bit about your company; the company name, RiverSilica, how you got founded and what you do?
Balaji: One of the things is, the company was founded in Bangalore and the name comes from a very metaphoric relationship to water. And water actually is something that takes the shape of a solid that it can be put into; so from that perspective it kind of connotates that we are moving towards streaming and the river and sort of what I call as a metaphoric relationship and the silica is basically the poetic justice to silicon. So that’s basically the background of the name RiverSilica and interestingly, one of the things that we probably are very close to looking to build a product that is closer to streaming, which is what I’m going to talk about later, is because India, probably-- where we come from and the company where we are founded which is in Bangalore, is actually a place in India which is as close or as equal to Hollywood so there is so much of content available, there is so much of video information that goes on I guess, probably, we are well suited to be in this kind of marketplace. So, that’s to start off as an introduction.
Siglin: Very good. I’ve spent some time in Bangalore and Hyderabad, South Indian cinema, as well as Bollywood and I completely agree. Hollywood is big to us in the US, we export well, but just the mass amount of content that comes out Indian cinema is certainly significant. So you do transcoding, and with transcoding you’re obviously looking to move content into various different formats but as I understand from the name Silica also you started out with the idea of either an ASIC or an FPGA and yet you’ve transitioned a bit as well so can you tell us a little bit about that?
Balaji: Oh gladly. So, we basically started to do sort of an ASIC in an FPGA for transcoding to start with and then as we started putting things in perspective we did quite a bit of analysis and we are pretty close to building an FPGA for transcoding, and I must say very close too, and we started looking at the costs per stream and also the flexibility required to build these particular formats that are coming out and with these two things actually we felt that we were at a bit of a disadvantage. And then we started looking to see how we could harness the algorithms that we had built to create high performance transcoding and we moved it to a general purpose processor or GPP we found that we could get better advantage. So that’s the kind of transition that we did. We have actually--, you know, we have seen one world, which is on the ASIC side. We are seeing the GPP side of it. I guess probably we are in a better position to tell you which one works better.
Siglin: So in the industry of transcoding, we have general purpose processors, as you talked about, you know, the standard CPUs. Then we have DSPs that some companies are focused on. We have GPUs that other companies have focused on. And then we have, on the high end, kind of ASICs and FPGAs. For you, in doing the work with general purpose processing, do you feel that you have an advantage over systems that say claim faster speeds or better quality because they use a GPU as opposed to a CPU or do you do a combination of both of those?
Balaji: Well, right now we actually do quite a lot on GPP and we might look at a GPU at some point of time but, you know, whatever we are doing right now we have optimized our software and we have added a lot of top up algorithms and then those top up algorithms are really serving us the purpose and if you want to add a GPU it would be an icing on the cake for us, right now.
Siglin: Oh, okay.
Balaji: So that’s exactly how we look at it right now and one of the things that I see is that transcoding is entirely a different perspective compared to encoding…
Balaji: …and the two are two distinct elements. So if you have to do transcoding it needs a different kind of perspective. You also need a bit of flexibility and also data that you can get from decode that you can analyze and then create an advantage when you take it to encode process basically. So to do all of that you need to have a mechanism that suits that kind of smart transcoding as opposed to pure encoding.
Siglin: So let’s talk about some of the things that your company claims: fast transcoding; lower price; higher quality. And one of the things you told me before we started was a lot of that is around the algorithms and you alluded to the algorithms here just a moment ago. Talk a little bit about what you think is the secret sauce for the company in terms of those algorithms.
Balaji: Well, I mean, it’s pretty interesting stuff. We actually spent nearly three years in trying to put together a bunch of algorithms, what we want to call it as tool sets basically, and these tool sets help us to harness a different kind of performance, extract different kind of, you know, advantages when we try to do transcoding; and transcoding, essentially, need not have to get into one type of application. You’ve got offline. You’ve got online. You’ve got purely online but you want to manually manage the bitrates, or you want to do online but you want to manage the frame rates, for example; so you’ve got all these parameters that keep going on. So, to do that, what one needs is some kind of a technique or an algorithmic technique that sort of takes the information about decode and encode and then figure out how exactly we can manage or manipulate those information to get where we want to go. So these algorithms, basically, are, you know, we are filing close to about 11 patents in this area and largely it comes from our brand names are RAFT, which is actually our RiverSilica’s Algorithms for Transcoding and algorithms based on SAN for scene detection or algorithms for H-E-L-M or what we call as HELM; these algorithms basically if you notice they’re all very nautical terms…
Siglin: Right, exactly.
Balaji: …so these algorithms basically help us to extract information, build these tool sets.
Siglin: Well the good news is with all the nautical terms because Inlet is no longer in the business obviously having been acquired by Cisco…
Siglin: …no one will confuse your product with theirs, potentially. So, you mentioned too that you had actually started out at IBM. Tell me a little bit about your kind of personal journey to how you’ve gotten to this particular company.
Balaji: Oh excellent. I mean, thanks for asking and I’m a typical middle class boy, average Joe, and I started off building symmetric multiprocessor systems for an Indian company called Vipra. And I moved on from Vipra to IBM. I built storage processors, did product marketing, worked in multiple geographies, worked for Dell for a while and then in 2000, 1999 actually I tried to have a startup with a friend of mine which went bust. Didn’t have any money, was on the…
Siglin: You had the perfect timing in 1999 to have a startup with the bust coming along.
Balaji: Actually, you know what, I mean, it’s pretty interesting in 1999 anybody who had actually a telecom background did great and we also had a telecom background and interestingly, you know, we had a lot of partners and I was a young kid then. I mean, I was about 29, 30 and then I thought, you know, I could do well in this startup but I just couldn’t understand what was going on and by then, you know, we were down to the last penny. And yeah, I mean, I had a choice between having another startup or getting into a professional job back again but I chose another startup with in 2000 which got funded by City Ventures; and this startup, largely we started building some sort of IP-based design services for semiconductors and given that I had a background in semiconductors and in symmetric processing and stuff like that so we quickly grew up and we were a team of about 350 people out of India. We were pretty close to being number three and as an Indian company had a lot of stuff going on in Japan and US and that’s where we sort of got into building image processing processors, intellectual properties systems and so on and so forth; that’s when we got into the video side of things. I’ve got a bit of personal background. My dad actually used to be, for a while, wrote some scripts in movies.
Siglin: Oh, interesting.
Balaji: So, you know, both of them actually kind of got together and then felt that this is an area that I think is a video explosion that’ll happen and be here kind of stuff; so that’s sort of background where I come from.
Siglin: And that takes us back full circle to the fact that as we said there are premium content around the world, whether it’s in Europe, whether it’s in the US, whether it’s India; all of that content needs to be transcoded to multiple formats. One last question before we wrap up. Primary focus on H.264 adaptive bit rate and the like or do you also include WebM and some of the future-facing things like H.265 and VP9?
Balaji: What we actually focus on is we’re a Tru-Transcode company basically and the art of Tru-Transcode is our mantra. We’re actually not too concerned about the future standards because the way we have built the product is to make sure that any standards that we move on we should be able to harness the advantage that we have created in the algorithm. So given that scenario, for us, HEVC is around the corner. We’re working on HEVC as well. Given that, I think from a product perspective we can actually put together a secret sauce on top of the new product line as we go by; that’s just to answer your question. And secondly, we pretty much would like to be on the ball when it comes to newer standards, the publishing side of things or transcoding HEVC or input, for example, at 4K for example. So, you know, in that perspective, yes, we are working on it and we probably are a kind of company we will go when-- we wouldn’t announce a product for the sake of announcing but we will have the right things that customer wants and the timing has to be right from that perspective; so that’s one of the things that probably I strongly propagate internally as well.
Siglin: Okay, very good. Balaji, thank you for joining us and those of you who are watching this video when you see the video on StreamingMedia.com there will also be a small write up underneath it but look forward to a future article about RiverSilica as well as they launch some of their new products. Thank you and have a good day.
Balaji: Thanks Tim and appreciate it.