Proposal: Modular Multilingual Support Architecture for Speak AI

[DhruvDG07]

I’m exploring the multilingual support idea for GSoC and wanted to understand the current architecture.
From initial exploration, it seems language handling is not fully modular yet. I’m considering introducing a structured pipeline:

• Language selection layer (user-controlled)
• Speech recognition layer with multi-language support
• Translation layer for extensibility
  Additionally, starting with Hindi support could be a practical first step.
  Could you guide me on:

1. Where language handling is currently implemented?
2. Any preferred libraries or constraints for multilingual support?
   I’d love to align my proposal with the current direction of the project.
   Thanks!