This project is a custom-built interactive audio survey platform designed to gather large-scale speech data for training AI voice and speech recognition models. The application allows volunteers to record pronunciations of 1,000 dynamically generated words through a clean, web-based interface.
Key Features
-
Volunteer Registration & Login – Users can securely sign up and log in to participate.
-
Step-by-Step Recording Flow – Each page presents a word, with an option to record via microphone.
-
Preview Before Submission – Volunteers can listen to their recording before submitting.
-
Resume Functionality – Progress is saved, allowing participants to pause and resume.
-
Random Word Generation – Words are dynamically pulled from the database to ensure dataset diversity.
-
Secure Storage – All audio files are uploaded and stored on the backend for later AI training.
Technology Stack
-
Backend: PHP
-
Database: MySQL (storing words, user credentials, and audio metadata)
-
Frontend: Web-based interactive UI with microphone input and playback controls
-
Hosting/Deployment: Configured for PHP-MySQL environments with secure file upload support
Purpose
The app’s primary goal is to collect diverse pronunciation datasets from volunteers worldwide, enabling more accurate AI speech recognition systems. The project showcases expertise in custom web development, database integration, and handling multimedia input in the browser.
