Drop an audio file, get a video. With just an audio file in input & thanks to a completely new Algo dashboard driven by audio, with video editing made as simple as editing text, the Bloomberg team can now create shareable podcast videos in minutes without the need to open any editing software.

Podcasts are continuing to grow in popularity, but since they are audio-only content, they are still hard to share on social platforms. To solve that issue, Algo worked in partnership with the Bloomberg Podcasts team to create audio-driven podcast trailers, using Artificial Intelligence.

The data-viz studio developed a unique audiodriven Algo dashboard for this project, where users can upload an audio file, and get an automatic and accurate transcript using the IBM Watson Speech-to-Text Artificial Intelligence (with an accuracy of 98% for the English language & supporting over 15 languages), and then split into video modules of 1520 seconds based on semantics. The Algo team designed a layout that can fit the overall Bloomberg Podcast brand identity and can also work with 4 different design skins dedicated to their 4 biggest shows.

From the transcript, each scene is turned into one of the template modules: Title Speaker (with waveform) Media (video, image) Maps Charts. The user can then edit the video timeline by simply editing the text. By wrapping the text and creating a new line to create a new scene or removing a line to merge two scenes into one.

With just an audio file in input & thanks to a completely new Algo dashboard driven by audio, with video editing made as simple as editing text, the Bloomberg team can now create shareable podcast trailers in minutes without the need to open any video editing software.

Algo is a data-visualization studio specializing in video automation, running a platform that turns data into videos, at scale.