translator

Imported from Zoioey/translator (AGENTS.md). Install upstream with npx skills add Zoioey/translator. Copyright stays with the author.

Project Rules

Goal

Build a stable end-to-end speech translation demo for internship assessment.

Focus on:

• reliability
• clean architecture
• engineering quality
• simple UX

Avoid:

• over engineering
• unnecessary dependencies
• premature optimization

Tech Stack

Frontend:

• Next.js
• JavaScript
• Tailwind CSS

Backend:

• FastAPI

AI:

• Whisper
• Google Translate
• Edge-TTS

Architecture

• Frontend and backend must stay separated
• Never call AI APIs directly inside React pages
• Use service modules for ASR / MT / TTS
• Keep components reusable and small

Suggested structure:

frontend/

• app/
• components/
• services/
• hooks/

backend/

• api/
• services/
• utils/

Workflow

• Explain implementation plan first
• Implement one module at a time
• Test after every module
• Never modify unrelated files
• Do not add dependencies unless necessary

Recommended order:

1. Audio upload
2. Whisper ASR
3. Translation
4. TTS
5. UI optimization

UI

• Minimal clean UI
• Show loading state
• Show logs for every AI stage
• Support audio playback
• Keep layout simple and readable

Coding Rules

• Use async/await
• Keep functions small
• Avoid deeply nested logic
• Reuse components when possible
• Prefer simple solutions over complex abstractions

Project Scope

This is an internship-level AI demo project.

Prioritize:

• stability
• completion
• usability
• clean structure

Do not:

• over design architecture
• introduce microservices
• add unnecessary abstractions
• optimize prematurely