Upgrade AI English Tester: Listening Module, CEFR Evaluation

Finalize AI English Tester — Add Listening Module & CEFR Evaluation (TypeScript + Vercel + Gemini AI + Supabase + ElevenLabs)

Description:
We already have a working MVP built with TypeScript (Next.js), hosted on Vercel, and integrated with Google Gemini AI, Supabase, and ElevenLabs.
The product is an AI-based English proficiency tester that conducts online interviews (Speaking, Writing) and evaluates users’ language level according to the CEFR (A1–C2) standard.

This is a project improvement task, not a new development from scratch.

Tasks:
Implement the Listening module (audio playback + questions + evaluation).
Build the final CEFR evaluation logic, combining results from all modules.
Integrate results into Supabase.
Perform stability and validation testing.

Deliverables:
Functional Listening module.
Valid CEFR-based scoring system.
Bug-free and stable MVP.

Preferred experience:
Google Gemini AI / OpenAI API integrations.
Supabase, Vercel, ElevenLabs API.
Experience with CEFR or AI-based language assessment.