Completed

Vocalis

Interactive pronunciation assessment tool for English, Chinese, and Filipino

System Architecture Visual

Project Overview

A pronunciation assessment tool designed to help users learn and practice pronunciation (supporting English, Chinese, and Filipino) by reading short texts and engaging in guided conversations. It leverages speech-to-text to evaluate acoustic inputs and text-to-speech to provide accurate reference feedback.

Methodology & Architecture

1. Audio Capture & Input Processing

Capturing live voice inputs in React Native as players read short texts or conversational prompts in English, Chinese, or Filipino.

2. Speech-to-Text & Transcribing

Transcribing speech inputs using speech-to-text to perform comparative analysis and verify accuracy against reference sentences.

3. Text-to-Speech Demonstration

Generating native pronunciation audios dynamically using text-to-speech engine to guide users with acoustic reinforcement.

4. Backend Integration (Spring Boot)

Managing backend logic, storing user scores, and tracking language-learning progress using Spring Boot with a Java database integration.

Technologies Used

Spring Boot
Java
React Native
Text-to-Speech
Speech-to-Text

Key Learnings

  • Integrating cross-platform mobile Speech-to-Text and Text-to-Speech APIs in React Native.
  • Structuring multi-lingual pronunciation assessment logic for English, Chinese, and Filipino.
  • Building robust, high-performance REST APIs in Spring Boot for user profile management.
  • Designing clean user flows to display real-time pronunciation feedback.