LingoDect

A Language Detection (LID) library written in Python, able to detect language from text, image or audio content.

Tech Stack

Back-end: Python (numpy, pandas, etc.)

Server: Flask

Models: scikit-learn (sklearn), TensorFlow, Keras

Front-end: HTML, CSS (Bootstrap), vanilla JavaScript

Installation

Install LingoDect with pip

Clone this repository, and from the repository root run the following command:

  pip install -e .

You should now have lingodect installed as a library and can use it for development and testing.

Run Locally

Clone the project

  git clone https://github.com/hadarsharon/lingodect

Go to the project directory

  cd lingodect

Install dependencies

  pip install -r requirements.txt

Start the server

  python app.py

Alternatively, you can run commands via the CLI (run with the -h flag for help and information about available commands)

  python cli.py -h

Running Tests

To run tests, run the following command from the project root

  pytest

FAQ

What types of inputs does this library support?

The library currently supports all textual input that can be written directly to it (whether as a string via the CLI or in a text box via the web application GUI), or via plaintext files such as .txt files.

Audio (Speech) and Image (Handwriting) inputs are available using most common input formats (.wav, .flac, .jpg, .png etc.).

In case your input file format is not supported, .wav and .png are a safe bet, so you should convert your audio or image file to them, respectively.

How many languages does this library support for prediction purposes?

The library models are currently trained on over 100 languages, so there is a good chance whatever language you want to predict is part of the support languages.