Skip to content

Understands Emotions / Sentiments from Audio File / Live Speech / Text File

License

Notifications You must be signed in to change notification settings

karthagokul/sentimeter

Repository files navigation

Sentimeter

Alt text

The module is published in pypi (https://pypi.org/project/sentimeter/) . you can install it via

pip install sentimeter

Introduction

Sentimeter is all about identifiying the sentiments from a source . The source can be a text / audio . Below are some of the built in modules shipping along with the package .

Audio Module

Process the Emotions in a given audio file. Currently it supports only wav file . In future we will add more formats

Live Speech

Process the Emotions in a live speech

Telegram Mode

Sentimeter has a Telegram Bot, it can process the emotions in a Telegram chat

Text File

Process a text file and identifies the emotions

Youtube Video

Under Development

Incase if you would like to create own modules, you can do it by extending the BaseSource class like below

class OwnSentimeterSource(BaseSource):
    def __init__(self, engine) -> None:
        super().__init__(engine)
        pass

    def run(self):
        # The text is is the input to the engine
        self.engine.process(text)
        return True

    def on_event(self, results):
        # This overridden function will get you the results(map of emotions)
        print(results)

We can use multiple backends as well with this engine. I have created an experimental ChatGPT backend . But by default it uses a built in model to identify the emotions.

Class Diagram

Alt text

Package Diagram

Alt text

Usage

There are applications developed using sentimeter. Please refer apps folder in https://github.com/karthagokul/sentimeter

Todo

  • Documentation
  • Google Hangout Connector

Setup

Create a Virtual env

py -m venv virtual-env
.\virtual-env\Scripts\activate

Install All Dependencies

pip3 install .

About

Understands Emotions / Sentiments from Audio File / Live Speech / Text File

Resources

License

Stars

Watchers

Forks

Packages

No packages published