Inspiration
I spent all of Friday on YouTube, and I found a video with heavy post-edit censoring. That got me thinking if there was a way to automate that process.
What it does
It uses AI to look through and use context to censor a recording.
How we built it
I built this project step by step, using Python libraries to record and queue audio, then using API calls and Open AI whisper to transcribe that audio into small text segments, which I could then send to ChatGPT to judge the offensiveness of that audio segment.
Challenges we ran into
I mainly ran into the challenge of making the live censoring work. That took up most of the time on Saturday, but seeing as I did not have much time left, I had to make the executive decision to switch focus to a more doable related project idea.
Accomplishments that we're proud of
I am proud to have actually finished a somewhat finished project for a Hackathon, especially since I didn't know I was participating until Friday.
What we learned
I learned a lot about AI API calling and Python libraries around audio recording and playing
What's next for Live Audio Censoring
I might work on localizing Whisper so that I can separate my audio into distinct words for more accurate censoring.
Log in or sign up for Devpost to join the conversation.