Real-time audio → text → translation → smart reply system built with Web Audio API, AudioWorklet, and OpenAI Whisper/gpt-4o. This project demonstrates AI-assisted live captions and contextual replies for online meetings.
This project captures live audio from system/tab using Web Audio API + AudioWorklet, processes it into 16kHz WAV, and sends it to OpenAI Whisper/gpt-4o for:
- 📝 Real-time transcription (captions)
- 🌍 Instant translation
- 💬 Context-aware suggested replies
Perfect for meetings, online classes, or live caption demos.
- Frontend: HTML, CSS, JavaScript
- Audio Processing: Web Audio API, AudioWorklet (resample 16kHz PCM WAV)
- AI Integration: OpenAI Whisper (
whisper-1/gpt-4o-mini-transcribe) + GPT-4o for translation & replies
- 🎤 Capture system/tab audio
- 📝 Real-time captions
- 🌍 Auto translation (EN ↔ ID)
- 💬 AI-generated suggested replies (context-aware & stable)
- 🔒 Lock mode: replies only refresh per new question
- Clone this repo:
git clone https://github.com/AndreasAle/Web-Audio-API-AI-Integration.git cd Web-Audio-API-AI-Integration