Spaces:

---

HK0712

/

FYP_ASR_Service

like 0

Running

Hello Hugging Face Team,

This project is a Pronunciation Analysis Tool designed to support two key groups: language learners and individuals with Speech-Language Pathology (SLP) needs.

Project Goal & Innovation

The tool's mission is to help users improve their speech by providing a unique "Speech -> IPA -> AI Analysis" feedback loop. This workflow is particularly valuable because:

• It addresses a global shortage of teachers and SLP professionals.
• There are currently no existing free tools that offer this specific, powerful analysis process.

Why a GPU is Essential

A GPU is critical for this project to scale and serve the community effectively.

• Concurrency: While the current CPU can handle a single user, it cannot process multiple requests concurrently. This means if two users try the tool at the same time, one will face a long wait or a timeout.
• Future-Proofing: To support more users and improve accuracy, I plan to upgrade to a more powerful model (like Coqui TTS's XTTS). A GPU is required to handle the concurrent load and deliver the real-time experience this tool promises.

Granting GPU access is the key to transforming this from a single-user demo into a robust, multi-user service for the community.

How to Use the Demo

Live Frontend Demo: https://pronunciation-analyzer-vue.pages.dev/

1. ASR Service URL: Please copy and paste the backend URL into this field: https://hk0712-fyp-asr-service.hf.space
2. Language Code: Enter one of the currently supported language codes: en_us or fr_fr.
3. Target Sentence: Type a sentence in the corresponding language (e.g., "hello world" for en_us).
4. Start Recording: Click to record your voice, then click again to stop.
5. The analysis will appear shortly after.

Thank you!