Voice Assistant/Anti-Earrape Bot

A Discord bot with two features: A simple voice assistant you can talk to even over others in the call and an ML-based earrape detection for an instant server mute to avoid ear pain.

Comment

Inspiration

Friend with a broken headset that chooses to earrape at random moments

What it does

  • Voice recognition - Google
  • Noise recognition - searching for high volume in sound waves, Tensorflow - we tried to implement, got stuck on conversion between numpy array and Tensor objects
  • Muting Discord users
  • Discord commands - to interact with the bot - code from GitHub repo - songs are searched online and added to queue, bot doesn't play the music - we tried to fix with FFmpeg, used multiple bots from GitHub (both in Python and JS, youtube_dl library was probably causing problems)

How we built it

We started in Python, switched to JavaScript and then returned to Python. First we dealt with discord bot commands, then voice recognition and finally noise recognition.

Challenges we ran into

async await - working with 2 libraries, one of them synchronous, the other uses async await FFmpeg setup Tensorflow conversion from numpy

Accomplishments that we're proud of

we can mute you when your mic explodes and therefore save people's lives we can use speech to text to control the bot (no need to click on discord to tell the bot to play a new song)

What we learned

Tensorflow, asyncio, pycord, Sphinx is bad, Google Speech to Text is good basic concepts of Fourier transforms

What's next for Voice Assistant/Anti-Earrape Bot

  • add more functionality
  • fix bugs to use current code

Built With

Share this project:

Updates