CSE498, Collaborative Design, Spring 2018
Computer Science and Engineering
Michigan State University

Amazon is an e-commerce and cloud computing company based in Seattle. Accounting for 1 in 3 online shopping transactions in North America, Amazon is the world’s largest online retailer.

In the third quarter of 2017, Nielsen reported that 40% of Americans listened to a podcast. Despite having a large audience, podcast producers have few opportunities to profit from their work.

Our Amazon Marketplace Podcast Earnings Detection system (AMPED) provides podcast producers an additional source of revenue by finding products related to their content. The results are displayed on a dashboard for producers with an audio player integrated with the items.

For example, if a podcast producer is doing a tech review on the latest MacBook Pro, AMPED searches the Amazon marketplace for the latest MacBook Pro, and provides an image and link for the listener to buy it.

AMPED finds relevant products by first transcribing the podcast’s audio content to text. Next AMPED comprehends the meaning of the text. Finally, AMPED searches the Amazon marketplace for the most relevant products.

Products appear in the bottom of the app as the podcaster mentions them. Listeners can choose to purchase the products in Amazon’s app or website by clicking on the picture of the product.

AMPED is built with Python and is hosted on AWS Lambda serverless architecture. AMPED uses Amazon Transcribe for speech to text, Amazon Comprehend for natural language processing, and Product Advertising API to find products.