[![Heroku][heroku-shield]][deployment-url] [![MIT License][license-shield]][license-url]

Misophonia Bot

A telegram bot to help you figure out if voice and audio messages are safe-to-listen

## Table of Contents * [About the Project](#about-the-project) * [Built With](#built-with) * [How it works](#how-it-works) * [Usage](#usage) * [License](#license) * [Contact](#contact) * [Acknowledgements](#acknowledgements) ## About The Project

[WebMB describes misophonia](https://www.webmd.com/mental-health/what-is-misophonia#1) as *"(...) a disorder in which certain sounds trigger emotional or physiological responses that some might perceive as unreasonable given the circumstance. Those who have misophonia might describe it as when a sound 'drives you crazy.' "* This bot aims to help people suffering from this condition judge whether or not a voice message or audio file is safe-to-listen. This all started as a joke between some friends, one of whom suffers from misophonia. Shout out to [@jvmcarneiro](https://github.com/jvmcarneiro) for the original idea. ---------- ### Built With * [Audio Class data annotation tool](https://github.com/glefundes/audio-class) * [PyTorch/Torchaudio](https://pytorch.org/) * [python-telegram-bot](https://github.com/python-telegram-bot/python-telegram-bot) ### How it works **Misophonia Bot** listens to any audio or voice message you send him and gives you his best estimate on whether or not that message is safe to listen to. Along with a verbal response with his recommendation, he gives you a score on a scale of 0 to 1 on where he thinks the message falls (with 0 being a clean message, and 1 being a very unsafe sound) so you can make your own judgement. The brain of **Misophonia Bot** is a Recurrent Convolutional Neural Network trained on scrapped youtube videos and 1000+ manually annotated voice messages from telegram group chats. The model was inspired by [this repo](https://github.com/ksanjeevan/crnn-audio-classification) and the training code will be available soon. Rest assured that no personal data is *ever* logged or stored by the bot. You can verify that by checking the [source code](https://github.com/glefundes/misophonia-bot/blob/master/bot.py) yourself. **Misophonia Bot** is currently hosted for free on Heroku, so some delays in response can be expected should the demand rise. ## Usage Record or forward a voice message to [Misophonia Bot](https://t.me/misophonia_bot) and you'll get your response. It's that simple! :) ## License Distributed under the MIT License. See `LICENSE` for more information. ## Contact Gabriel Lefundes Vieira - lefundes.gabriel@gmail.com ## Acknowledgements * [README Template](https://github.com/othneildrew/Best-README-Template) * [CRNN for audio classification](https://github.com/ksanjeevan/crnn-audio-classification) * [Skeleton for app deployment of python-telegram-bot on Heroku](https://github.com/Eldinnie/ptb-heroku-skeleton) [stars-shield]: https://img.shields.io/github/stars/github_username/repo.svg?style=flat-squarehttps://img.shields.io/appveyor/build/glefundes/misophonia-botz [stars-url]: https://github.com/github_username/repo/stargazers [license-shield]: https://img.shields.io/github/license/glefundes/misophonia-bot.svg?style=flat-square [license-url]: https://github.com/glefundes/misophonia-bot/blob/master/LICENSE [heroku-shield]:https://img.shields.io/badge/Heroku-deployed-blueviolet?style=flat-square&logo=heroku [deployment-url]:https://github.com/glefundes/misophonia-bot/deployments