Skip to content

About

This project uses an HMM model to convert Romanized Sinhala names into phonemes. This improves Google TTS accuracy, allowing it to pronounce Sinhala names more naturally and correctly. It significantly enhances the user experience by providing authentic pronunciations.

Resources

Stars

1 star

Watchers

0 watching

Forks

Repository files navigation

IT5514 Final Project

Improving the pronounciation of Sri Lankan cities and names.

Methods

  • Hidden Markov Models
  • Transformers

Dataset

  • Obtain a list of names
  • Convert them into International Phonetic Language

Data Collection Tool Features

Download Pronunciations

The data collection tool now supports downloading all pronunciation data as JSON files with the following options:

  • My Pronunciations: Download only your pronunciations
  • Valid Pronunciations: Download all valid pronunciations (marked as valid by users)
  • All Pronunciations: Download all pronunciations in the database

Access Methods

  1. Navigation Bar Dropdown: Click the "Download" button in the top navigation bar for easy access
  2. History Panel (Removed): Download buttons previously in the history panel have been moved to the navigation

Features

  • Visual Feedback: Loading animations and status messages during download
  • Smart Dropdown: Click-to-open dropdown with outside-click-to-close functionality
  • Mobile Responsive: Optimized dropdown positioning for mobile devices
  • Rich Icons: Visual indicators for each download type (user, verified, database)

JSON Export Format

{
  "metadata": {
    "export_timestamp": "2025-07-12T10:30:00.000Z",
    "filter_type": "all",
    "total_records": 150,
    "exported_by": "user_uid"
  },
  "pronunciations": [
    {
      "id": "doc_id",
      "city": "Colombo",
      "pronunciation": "kəˈlʌmboʊ",
      "is_valid": true,
      "user": "user_uid",
      "user_name": "John Doe",
      "timestamp": "2025-07-12T10:15:00.000Z",
      "created_at": "2025-07-12T10:15:00.000Z",
      "updated_at": "2025-07-12T10:15:00.000Z"
    }
  ]
}

The download feature is accessible from the History panel in the data collection interface.

About

This project uses an HMM model to convert Romanized Sinhala names into phonemes. This improves Google TTS accuracy, allowing it to pronounce Sinhala names more naturally and correctly. It significantly enhances the user experience by providing authentic pronunciations.

Resources

Stars

1 star

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages