Qualcomm® Developer Network February Developer of the Month is our AI Developer Contest grand prize winner Sergey Konvisarov from Espoo, Finland. His Pencil Music App will generate different styles of music based on user drawings in real time.
Sergey has been working with computers for over 30 years spanning the birth of MS-DOS, Basic, Connect 2400 MNP5, FidoNet, Internet, and Google’s triumph over AltaVista. He enjoys working on hobby projects (including his Airyware Tuner app) and learning new things, especially when they’re very difficult problems. This explains his current fascination with Artificial Intelligence.
What do you like about working with Artificial Intelligence (AI)?
I really like the fact that AI can do things that were previously impossible. Recent achievements in the field of AI are awesome. It is especially cool to be a part of the trend, or at least pretending to be a part of the trend. ;-)
When taking part in the AI Developer Challenge, how did you approach your work process?
While it was easy to come up with the idea it was hard to implement it. First, I prioritized the critical features and then drew a simple architecture of the module interconnections (synth, composer, camera, AI, draw pad). Then I started coding. To speed up the process I worked on each module separately.
There were lots of challenges and each time I found a solution I could feel the fire in my belly push me forward. As the deadline approached I married all parts together and started shaping it into a demo version. Although I wanted to keep working on it, at some point you have to say to yourself “Stop. No more new features. Stabilize and finalize what you already have.” I wish I had cut myself off one day earlier because shooting and editing the demo video on submission day was not super fun.
How was your experience working with Qualcomm Snapdragon™ Neural Processing Engine (NPE) SDK and the Qualcomm Snapdragon 835 HDK?
The documentation and available samples were helpful to quickly start using the SDK. I really enjoyed the performance of running a neural network on the GPU. Qualcomm Technologies engineers did a great job providing this functionality.
While my app did not require any external peripheries except camera, it would not have been easy to debug and profile the app on a consumer device without rooting it first. With a dedicated Snapdragon 835 HDK it was a real pleasure to debug and optimize app performance.
In short, without these two technologies the Pencil Music app would not have come to life.
How do you plan to use products from Qualcomm Technologies in your next development project?
I have an idea of fusing the core engine of Airyware Tuner with a deep neural network running on Snapdragon NPE SDK to recognize the instrument being played and automatically associate the corresponding tuning profile, frequency range and sets of analyzed harmonics with their AGC envelopes. Theoretically, it may also help to tune in very noisy environments where it is not sufficient to use the standard noise cancelling feature of Airyware Tuner.
Where do you see Artificial Intelligence in 10 years?
In 10 years, while casually engaging in small talk with your home robot, AI could artificially limit or boost its own intellect to match your IQ level to move to laughter or to teach you something. AI will be everywhere: in transport, medicine, entertainment, and finance. You will ask AI “where do you see humans in 10 years” and I hope the answer will not depress you! ;-)