[Developer Interview] Q&A with Leslie Spring, Creator of SILVIA for Android
I had the opportunity to interview Leslie Spring, the CEO and creator of SILVIA for Android. We at OMG!Droid, have been following the team since we first saw the start up over at Kickstarter.com. After speaking to Leslie about his development background and learning more about the details of app, I am sure that SILVIA is going to take the Google Play Store by storm. In my opinion, the other voice systems out there won’t be able to hold a candle to SILVIA once it’s released. So without further adieu, let’s get into the interview.
Rayman: I understand that you had a hand in the development of commercial products with Disney, Electronic Arts and Sony? Can you tell us about some of the products you have help developed?
Leslie: My work was never on the product itself, general game and tools engines that game developers used. I did work with Disney on a 3D Aladdin game, I also worked directly on a few sports titles during my time with Electronic Arts, but “That was long ago in a galaxy far far away”. (First mention of Star Wars)
Rayman: Why did you get involved in developing SILVIA? What brought you to where you are now?
Leslie: It has been a long and winding road since my mom took me to see ’2001: A Space Odyssey’, but the light really went on for me when I saw ‘Star Wars’ (yes OMGer’s he is a Star Wars fan), this is what really made me make the decision to get into programming. I felt that the use of computers on movies like Star Wars (Yes this is the third mention) was just the beginning. I started with computer programming and then moved over to video games and building art tools for game designers, and then moved into 3D animation tools. Over the years I really got into the architecture and tools that helped the game developers design and build better faster games.
About 5 years ago I had a eureka moment as if I was Doc Brown from ‘Back to the Future’ slipping and hitting his head on the sink and creating the Flux capacitor (Not Star Wars, but a great reference). I had been thinking for a long time about how I should approach this problem: intuitive software that appears to have thoughts of its own. I was going to do this by separating the human language processing from the machines conceptual thought process. I found a way to decouple the two processes and isolate each given process, on the front-end I could have it be unique and interesting, and on the back-end I could work on the actual machine conceptual thought process that would make up SILVIA.
Rayman: What made you decide to take SILVIA to the Android platform?
Leslie: The run time is very cross-platform and we have applications that are deployed on the windows and a few other mobile devices behind the screens. We also do work with a major defense company that uses our artificial intelligence solution for training not available for public consumption.
Coming out in terms of mobile applications Android has provided us with a much more open platform, a greater market penetration (Android is taking over), we also feel we can deliver an easily-used and compelling platform. Let me give you an example. In the past, most of the demonstrations we have done on a PC but now most of the demos are being completing from the mobile device running SILVIA for Android. This has allowed us to make a very compelling demonstration.
Rayman: What’s going to distinguish this from other voice programs?
Leslie: SILVIA will be 7mb at launch on the device. Yes that is everything folks I am guessing it will be the smallest voice program out there. It will grow in size as it gathers the historical context, but we have special conceptual compression algorithms that we are using that turn your conversational history into a tokenized format, which is accessible by the SILVIA system. There are three big components to why we use this software. First, it takes up very little space. Second, it’s easily decompressed into its original form and third, it is secure and encrypted.
SILVIA will also store your conversational historic context, no server side support as it will not be needed for SILVIA to run. SILVIA will also learn based on your inputs both voice and physical use.
We do not do speech recognition nor do they do text-to-speech, we are the brains behind the scene. We are using Google speech recognition which in improving by leaps and bounds from all the user data they are gathering from users over the years and is freely available. Other platforms do not have that to offer and the cost involved with getting that information would not allow them to provide the best experience to the user.
Lastly, security is strongly encrypted and no one has access to the data unless you give them access. As an example “SIRI” is not allowed in IBM, because the conversational data is not stored locally unlike the conversational data from SILVIA that is all stored locally on the device and not shared over a server.
Rayman: Devices that did not start with Froyo 2.2, will they still be able to run SILVIA?
Leslie: SILVIA will be able to run on any device that has Froyo 2.2 and with so many devices out that that run android we will continue to test the devices and work with the manufactures to ensure the end-user experience is what it needs to be.
Rayman: What made you decide to include a 3D version for the gamers out there?
Leslie: Imagine an MMO and you are a fantasy character and interacting on line and imagine that you would not be able to tell if you where interacting with an actual person or SILVIA. SILVIA learns you as a player, learns how you speak and play. SILVIA can actually take samples and produce new output to interact with people on your behalf. We have unique algorithms for language output that can synthesize new output on its own.
Rayman: There are other voice software apps that claim to run natively. Will SILVIA need to be prompted to interact?
Leslie: There will be options, some users may want to run SILVIA,interact and then exit. But there is also an option for you to have SILVIA always listening and prompting by saying “Hey SILVIA”, and then she would respond to you. One other way is an interface element, but all of these options are on the table and still being worked. We want to provide users the options to choose what they want SILVIA to do.
The app will have three different core personalities that will be bundled on launch. Secondly the personalization kit will be able to be used by developers and the non-developers to allow for AI customization and a direct link will, be provided between the personalization kit and your phone to make it very easy for everyone to customize how your SILVIA will interact and speak to you.
Rayman: Do you think you will reach the target of 150,000 with only 15 days left?
Leslie: With 15 days left, we are shifting the target a little and targeting the developers a little more. We are also working with a team on social networking, and with interviews like this and the one from Bloomberg, we are remaining positive and have had a great experience with Kickstarter.com. One of those reasons is that we are getting an incredible amount of positive feedback from the community that is helping to drive the project and have consumers tell us what they want to see.
For your viewing pleasure…This interview really opened my eyes to the brains behind the scene at Cognitive Code. With SILVIA launching at a $20.00 price point, this has been a great experience and I want to thank the Leslie Spring and the team for allowing OMG!Droid the opportunity to interview.
Company Site : Cognitive Code