We give up. We are too tired :(

Inspiration

The loud background noise of the hackathon made it hard to understand each other.

What it does

Using microphone data from multiple users' phones, this project cancels localized background noises. For speech, an additional logMMSE layer is used to enhance the sound.

How we built it

Each phone runs an app that sends microphone data to the server.

Challenges we ran into

A power outage, modules and APIs not working together, and the general challenge of denoising and enhancing speech.

Accomplishments that we're proud of

What we learned

Speech enhancement is hard.

What's next for BigMic

Streaming, real-time noise canceling. Speech recognition. Machine-learning based speech enhancement.

Share this project:

Updates