🥇 Best Overall - Hack GT 8

🥇 Best Design - Hack GT 8

What is Hypercut?

Watch what matters most

Hypercut is an app that makes your videos shorter. It extracts features from the audio track to identify pauses and prioritize phrases. Users can control how to cut the video, from slightly speeding up pauses, all the way down to a two-sentence summary.

Why would I use Hypercut?

It's easy to lose your audience's attention, especially with non-interactive videos. Hypercut can keep your audience interested by livening the pace of the video or even finding the most interesting parts of the video.

Hypercut could be used on lecture videos to make quick crash-course style reviews. Hypercut is also great for building a highlight reel of business announcements, conference proceedings, or news. This is especially relevant for making quick 30 second videos for social media.

Fancy... How does the tech work?

The app begins by uploading an audio file to the Hypercut backend for feature extraction. Hypercut uses speech-to-text to identify timestamps of words and pauses. Using a modified implementation PageRank, Hypercut can assign a priority value to phrases.

The feature extraction information is sent back to the app, where the user can drag sliders, forming exclusion criteria for the features. Finally, the app encodes a new video from the selected features.

How did you make that UI?!

The design is inspired by the principles of neomorphism and magnetism. These principles help build tactile and responsive software interfaces.

The interface is built with custom SwiftUI components. They are built by stacking lots of rectangles and text boxes with custom computational drag-resistant micro animations. Cool right?

What's next for Hypercut

Hypercut is only a few steps away from detecting sonifications like "ummm" and "like". iPad support is also a near goal. A little more into the future, we believe Hypercut could provide a suite of tools for live audio, like in Zoom calls, where it could remove "ummm" and "like" on the spot! With a touch of neural voice puppetry, we see Hypercut being able to provide a virtual camera to users, masking filler speech and unwanted visual disruptions to provide the best, most professional live presentation possible!

And that's not all. Hypercut can be further extended to help students and professionals review material, better and faster. We already have word-by-word timestamps, which can provide Hypercut users a way to quickly search through videos by text. Hypercut also finds the most relevant information in presentations, allowing us to provide video summaries in both text and video.

Built With

Share this project: