Insights / Industry Perspectives / Case study: The AI search engine for podcasts is a media differentiator

·

4 mins read

Case study: The AI search engine for podcasts is a media differentiator

Today’s tech-minded audience is in full control of what content they want to consume, when they want to consume it, and how.

Media consumption became even more interesting during and after the pandemic when giving customers a supremely personalized experience that would cater to their unique needs became critical. At the same time, next-generation technologies like AI (artificial intelligence), ML (machine learning), NLP (natural language processing), and computational creativity have turned traditional media and entertainment providers upside down.  

Here’s the fact: Media and entertainment companies can no longer create one-size-fits-all solutions, and giving customers a highly personalized experience has become imperative. Using technologies like machine learning, companies can quickly identify podcast content topics with more precision and efficiency.

With so much content being produced daily, people do not have time to spend hours searching for content that matches their needs. Luckily, NLP can filter unstructured and structured data and instantly provide users with the right segments.  

This helps users search for important micro-moments that add value to their personalized viewing experience. It also opens a world of opportunities for advertising companies to deliver the right content and drive deeper engagement with audiences by matching the right ad with the right demographic.  

There is a huge gap in the podcasting space that prevents growth and monetization of the medium: topics that people are interested in are buried inside long podcast episodes.

One of HTEC’s clients recognized that gap and saw it as an opportunity to build an AI search solution that would cater to users’ needs by giving them just the right amount of content they want to hear. 

The opportunity: AI search is better for users and advertisers

The client approached us with the idea of creating an AI search solution that would allow users to find relevant topics in podcasts more easily without having to do it manually and/or listen to the entire podcast.  

Value created: An AI search engine for very specific podcast topics

HTEC helped the client build an algorithm, a neural network for segmentation, which used machine learning to analyze podcast content, identify topics discussed in the show, and surface those topical hooks as podcast search engine results.

At its core, the solution cuts podcasts into smaller audio segments, allowing users to easily find only the topics they want. Now, the user does not have to search through the entire episode manually because the algorithm gives users just what they need within seconds. 

For instance, a travel podcast covering the topic of green energy may include a discussion about different solutions that would make the world we live in more environmentally friendly. If a listener wants to hear more about the digital plants of the future, this AI-powered podcast search engine algorithm can help them find that (otherwise hidden) topic in the podcast right away. These micro-segments empower listeners to focus on their areas of interest while, in parallel, providing a platform for advertisers to position their content closer to these interest areas. 

In addition, HTEC created a web platform where users can participate in building audio segments using the algorithm, allowing them to foster a social discovery environment and encourage viral listening. 

The core of the project was scientific research rather than a straightforward implementation, as we explored different architectural solutions for neural networks to find the best ones to help us build our algorithm. 

Challenges: Search algorithms are only as good as their databases

When building this kind of algorithm, having a good database to train and test algorithms is critical. The higher the quality of data, the better the algorithm.

To achieve that, the HTEC team spent a year exploring the different podcasts the client selected, which provided enough information for the team to start building a tool for labeling data.

The team created the tool and built a huge database for algorithm training fueled by NLP technology. The team ended up with a core algorithm that beat most of the competitors on the market. The so-called “Schreder” was almost 80% successful in cutting and selecting segments based on the database that HTEC created.  

The next step was to create a more artistic image representation of those audio segments (the topics people want to listen to with all the keywords). The team decided to create an algorithm pipeline with a full episode or a series of many episodes on the input and audio signal with an image representation of the topic, the speakers in the podcast, keywords, and a summary of the topic discussed on the output.

The team selected a few algorithms and implemented the pipeline. We finally built a fully automated audio segment with a visual representation that listeners could listen to with a click of a button. From a UX perspective, each audio segment has a visual representation (collage and cinematography) supported by NLP and AI. 

Impact: A win for content creators, advertisers, and consumers 

While listeners/viewers can move faster through multiple podcast episodes, consuming curated content based on their preferences, the content creators are empowered by reaching more relevant audiences with the right material.

The synergy created via this new approach signals a new direction that could disrupt existing global advertising models. 

Want to learn more about how our technology expertise can transform your business? Explore our Product Engineering and Media & Entertainment capabilities.


Author