arrow_back Back to Works
Web Application

Automated Listening Test Generator

A web app that automatically generates listening test scripts, audio, and questions, with per-speaker accent and speed settings and an audio playback UI. Live in production.

Duration

Contract development

Team

1-2 people

Engineered With TypeScriptCloudflare Pages

01 // OVERVIEW

Project Overview

A web app that automatically generates English listening test scripts, audio, and questions in the browser. Users configure a monologue/dialogue format, CEFR level, speaker nationality, accent, and speed to generate a script, then synthesize audio via Google TTS. It includes per-segment blank-time settings, a playback UI, and folder management, and is live in production on Cloudflare Pages.

02 // KEY FEATURES

Key Features

hearing

Automatic script and question generation from settings

Specifying a topic, keywords, CEFR level, number of speakers (2-5), and each speaker's nationality automatically generates a listening script and multiple-choice questions. Scripts can be edited and regenerated directly in the browser.

record_voice_over

Per-speaker accent, speed, and gender settings

Each speaker's accent can be set from 7 regions, including American, British, and Australian English, along with gender and speed (0.5x-1.5x). The blank time between lines can also be adjusted individually.

folder

Folder management and audio playback UI

Generated tests are saved and managed by folder, with audio segments playable individually or continuously as a full track. Also supports WAV file download and QR code output.

03 // TECHNICAL DEEP-DIVE

Technical Highlights

Solution

Ensuring precise blank insertion and WAV concatenation

Required precisely decoding and concatenating blank seconds after each line using the Web Audio API. Fixed a bug where JSON responses were mistakenly saved as audio, and used displayIndex for accurate input mapping so the order and timing of every segment lined up correctly.

04 // IMPACT & RESULTS

Results

Live

Running in production on Cloudflare Pages

7 regions

Supports accent-specific speech synthesis

Fully automated

Generates script, questions, and audio in one pipeline

Ready to Build?

Want to build something similar? Tell us about your project.

Free Consultation