“AI Is at the Core of Pocket FM and Personalization Sits at the Top of it” Says Prateek Dixit

By S G Rickman On Nov 23, 2022

The very concept of having the time or not having it to rejuvenate oneself with good content has obsolete. Doom-scrolling to find the audio that will engage us for the next 30 or so minutes while we go around our everyday chores is passe. The audio on-demand platforms have evolved to users’ tastes so much so that one can go hands-free and still keep oneself satisfied and emerge even saner in the end. Thanks to artificial intelligence and recommendation systems, audio OTT platforms are finally serving the purpose. Prateek Dixit, in this exclusive interview with Analytics Insight, takes us around the world of AI and data-driven audio content delivery.

1.Kindly brief us about Pocket FM.

Pocket FM is the pioneer of the audio series category, a long-form audio storytelling concept. This is an entirely new category in the audio entertainment space that is built with a vision to redefine the audio entertainment OTT landscape. We have grown exponentially on the backdrop of AI-powered algorithms built in-house to create a better experience for its listeners, improvise our content delivery, and thereby, contribute to our bottom line

Over the years, the product has evolved into a leading audio streaming service on the back of its personalized content recommendation engine. We have been front-ending the category formation for the audio OTT landscape with the audio series. We have made significant efforts by building a community for audio content and, presently, dominate the non-music entertainment audio space in India.

We have pioneered the habit of binge-listening with long-form storytelling and engaging content created by our 500,000+ strong creator community. Today, Pocket FM witnesses over 100 minutes daily average listener time spent. The total monthly audio streaming on the app counts to over 3.5 billion minutes.

2.With what mission and objectives, the company was set up? In short, tell us about your journey since the company’s inception.

Conventionally, the audio content landscape has always been imposing standard content offerings on the listeners for consumption. This often resulted in irrelevance, annoyance, and irritation for the listeners. With digital interference, this has changed slightly, but storytelling and personalization of content remained the core challenge even then.

We started Pocket FM to solve these two core problems – build an effective storytelling approach and create a strong content recommendation engine to enable personalization on our platform. In the process, we established a newer category in the form of audio series that has been the core of our offerings.

We always focus on delivering storytelling on Pocket FM through engaging content, as it increases listeners’ stickiness on the platform and their appetite to pay for content.

3.Tell us how Pocket FM is contributing to the advancement of AI in India. And how is the company benefiting from its AI infrastructure?

With unique products, there come unique challenges too. No other audio platform has seen the scale it operates in the audio space, at least in India, if not globally. There were challenges in ensuring the quality of content and maintaining the audio standard, and we have adopted AI/ML to automate the processing of audio files at scale.

At Pocket FM, AI is about building great user experiences, system automation, and eliminating manual and redundant processes. With a community of over 500,000 creators creating content at scale, the company has prioritized automation along with the listener experience. The focus is to ensure the listeners have the best listening experience with the most relevant content, quality audio standards, and undisrupted audio streaming.

Further, Pocket FM does not believe in complicating the engineering process. Instead, we focus on delivering the best experience to our listeners. The philosophy revolves around its planning cycle, which starts with creating better experiences for the listeners and creators with a belief that experience leads to a better revenue path.

4.Kindly share your point of view on the current scenario of AI and its future.

While AI has been around for more than just a few years, nowadays, it has an exponentially growing influence on the development of our society, economy, and digital age. It’s so rapidly evolving and being deployed that we can see it changing how we live and interact. Artificial intelligence is shaping the future of humanity across industries. It is already the key driver of emerging technologies like big data, robotics, medical sciences, and IoT, and will continue to act as a technological innovator for the foreseeable future

5.Please tell us how your listeners are driving values from Pocket FM, through your AI algorithms

Personalization became necessary with content varying across genres like horror, thriller, fantasy, and romance, with its diverse listeners’ community, and various Indian languages.

We built deep learning AI models to create unique listener profiles based on their content affinity, which serves as the core backend for the personalized content recommendations engine. We have one of the best AI teams in the content landscape.

As most of India is still accessing the internet in 2G and 3G networks, we have ensured bandwidth optimization for a better user experience and content delivery at its highest quality. Being an audio platform, our content universe was dominated by millions of hours of audio content. With scale, Pocket FM faced the challenge of processing these audio files to extract the data for machine learning, and they built their deep learning process internally.

Our deep learning infrastructure takes these audio files and converts 20-30% of the content into text through our speech-to-text intelligent conversion. The exercise helps create various metadata, including tags, categories, genres, etc., for the audio files, which is then used by the data science and content pipeline to strengthen the recommendations engine further.

6.Please share some major achievements of the company under your leadership

Since it is a whole new category and community-driven long-form audio entertainment, it also faces challenges in ensuring the quality of content, maintaining the homogeneous audio standard, etc.

There are few specific solutions in the market for audio, and it continues to struggle for ad-hoc audio processing and moderation. Therefore, to address the problem, we have been building these solutions internally using AI/ML to automate the processing of audio files at scale. We are also building our in-house audio privacy systems, which should help us automate the quality checks pipeline. With the scale at which Pocket FM operates, it becomes very hard to manually moderate millions of audio files.

Being an AI-first organisation, we have been eradicating manual intervention as we develop newer algorithms to stimulate its automation process. It keeps improvising the system with data points from the listeners like network operators, internet speed, and the historical consumption pattern. These variables help build better listener profiles and bring a more customized approach that optimizes experiences.

7.What does your technology and business roadmap look like in the next few months?

AI is at the core of Pocket FM, and personalization sits at the top of it. We are further doubling down our efforts on our AI and ML capabilities and focusing on building an advanced personalized content recommendation engine to drive higher user engagement and retention.

Generative AI is another important focus area for us. We plan to invest in building robust generative AI capabilities across language translation, synthetic voice production, and text-to-visuals.

The post “AI Is at the Core of Pocket FM and Personalization Sits at the Top of it” Says Prateek Dixit appeared first on Analytics Insight.