Skip to content

Taha-Firoz/UrduNLPHelper

Repository files navigation

Urdu Audio Dataset Generation Helper

This tool is a simple gui that lets you view your urdu sentence and record it with ease, it stores each audio file witht he respective ID for the sentence. At the moment it's specifically made for NUCES NLP Urdu dataset file and there are no plans of expanding it.

Features

  • Record, Redo and Save audio files
  • Convert Urdu text to Roman Urdu for easier readability (Unstable)
  • Easy shortcuts like s key to record/redo/stop n key or d key to move to next sentence

Please run the following commands for windows installs:

You need the same version of chrome driver as your chrome browser version. Keep the chrome driver executable in the same directory.

pip install -r requirements.txt
pipwin install pyaudio

or if you want to install the libraries individually

pip install wave
pip install pipwin
pipwin install pyaudio
pip install selenium

To run the program

python app.py

Troubleshooting

Incase the program doesn't start recording your audio and gives and exception or error. It's most probably related to a sampling rate not being set to that pythons expecting.

  • Go into your microphones advanced settings
  • Choose the advanced tab and set your sampling rate to 44100

About

Basic urdu audio dataset creation tool

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages