In India, the video streaming industry is transitioning from a niche market of a millennial, young male from metros to a mainstream market with more than 60% viewership from non-metros.
Gopichand Katragadda, former Group CTO, Tata Sons, has donned the entrepreneurial hat. Along with two former Tata Sons colleagues (Aditi Olemann and Ganesh Suryanarayanan), he has set up a deep tech venture called Myelin Foundry in January 2019, which is building Artificial Intelligence (AI) models on edge devices (such as mobile phones, wearables, laptops) for complex unstructured data (such as video, audio, speech, sensor data). Myelin is building AI products that will enable industries such as media & entertainment (M&E), to provide ultra-premium quality of experience for viewers, and bring down cost and time to market for new content by one-tenth. The Myelin platform will also enable industries such as healthcare to provide AI driven non-invasive prediction of a person’s state of health and wellness. Katragadda talks to Sudhir Chowdhary about the technology Myelin is developing, which is a global first, and how the venture is transforming human experience and industry outcomes with AI and deep learning. Excerpts:
What was the reason behind starting Myelin Foundry?
My co-founders and I have had the benefit of global experience in large and small companies and delivered critical products both in India and in global markets. We wanted to leverage this exposure, experience, and network to create a deep tech company out of India with global-first products, leveraging AI at its core. The three of us have creative pursuits in visual arts, hence the focus on M&E as a domain.
Myelin Foundry is a deep tech product startup that transforms human experiences and industry outcomes by building AI algorithms on video, voice, and sensor data for edge devices.
What are the various solutions being offered by Myelin and what problems are you trying to solve?
Myelin’s immediate focus is to accelerate the M&E industry’s transition to over-the-top (OTT) streaming services. Our OTT product, Fovea AI, deploys on edge devices and enables the highest quality visual experience at the lowest bandwidth consumption for OTT providers and viewers.
Video streaming is not usually profitable, and the process is inefficient using engineering hacks such as adaptive bit rate streaming to overcome bandwidth issues. Today, the willingness to pay for streamed content is low, and, most content is still viewed in Standard Definition. Re-buffering is still a big challenge. Premium OTT viewers demand cable TV-alike experience with end-device capabilities outpacing content availability. 4K and 8K content is limited. Game streaming requires higher constant bandwidth for high viewer experience.
These challenges are set to increase with 5G. Fovea AI enables efficient video streaming while enhancing the viewer experience. We are also cognizant that every bit of a video stream results in climate impact. Fovea AI enables the lowest carbon footprint for streaming.
The Myelin team has expertise in productising deep learning frameworks including CNNs and GANs. Our R&D pipeline also has solutions for the health and wellness industry.
OTT space is a key area of focus. How does your solution impact the space?
A recent report forecasts the video streaming market to be more than $500 billion by 2024. We anticipate that with video streaming costs coming down, and end user experience improving, new opportunities for video streaming will emerge in areas such as corporate training, education, and live commerce. In India, the video streaming industry is transitioning from a niche market of a millennial, young male from metros to a mainstream market with more than 60% viewership from non-metros.
Myelin’s Fovea AI enables zero re-buffer OTT with the highest quality of experience on the device, at a fractional cost. The Myelin solution performs sub-pixel convolution and further processing, real-time and at the edge that enable deployment at 30 fps and even 60 fps, thus enabling both video and game streaming. Today our solution provides real-time video super resolution and artefact removal, greater than 30% reduction in average streaming cost including reduced transmission cost, reduced transcoding cost, and reduced storage cost. We also significantly enhanced viewer QoE providing device-aware consistent video quality (upto 4K), eliminating rebuffering (at minimum 200 kbps), and we eliminate bitrate switching. Fovea AI reduces in data usage for viewers allowing two times the number of HD videos at same (limited) data plan or using half the bandwidth for 4K quality video (for TV). Future opportunity includes dynamic range optimisation, content-aware video search, ad insertion, summarisation, logo, brand and face recognition and related call-to-action by learning, visual content, audio, style and other aspects of the video and posters.
How do you see your team expansion?
As part of our expansion plan, we will be looking to hire domain experts who will help us solve the right problems for the respective industries. We will be building a business development and marketing team to help us grow our markets locally and globally.
What are you plans for future expansion and funding?
In the M&E space, we will be launching Fovea AI in India early next year. Our expansion plans will first focus on India-like markets before launching the product for the global premium market. We will infuse Series A funds by May 2020.