1️⃣ PhonoCoach

Chrome extension with FastAPI backend for real-time pronunciation feedback using phoneme analysis leveraging OpenAI's whisper ASR model.

---

2️⃣ Getting Started

Prerequisites

• Python 3.12+
• pip
• Chrome browser
• ffmpeg installed and available in your system PATH.

# a)Clone the Repository
git clone https://github.com/Manoj-HV30/phonocoach.git
cd phonocoach
# b)Create a virtual environment
python3 -m venv venv

### c)Activate the virtual environment
# Linux/macOS:
source venv/bin/activate
# Windows (PowerShell):
# venv\Scripts\Activate.ps1

# d)Install dependencies
pip install -r requirements.txt

# e)Start the FastAPI backend server
uvicorn backend.server:app --reload

Load the Chrome Extension Locally

1. Open Chrome and go to chrome://extensions/
2. Enable Developer mode (toggle in the top-right)
3. Click Load unpacked and select the frontend folder inside the cloned repo
4. The PhonoCoach icon should appear in your toolbar

3️⃣ PhonoCoach in action

Using PhonoCoach

1. Select any text on a webpage.
2. Open the PhonoCoach popup.
3. Click 🎙 Record to start recording your voice.
4. Click Stop to upload audio and analyze pronunciation.
5. View similarity score, phoneme-level feedback, and improvement tips.

4️⃣ Features

• Real-time pronunciation analysis for any selected text on any webpage
• Phoneme-level feedback highlighting correct, incorrect, missing, and extra sounds
• Similarity score to quantify pronunciation accuracy
• improvement tips based on your performance
• Uses OpenAI's Whisper ASR for accurate speech-to-text transcription
• Lightweight FastAPI backend for fast processing

5️⃣ Dev Notes

• Make sure the backend server is running before using the Chrome extension.
• Ensure ffmpeg is installed and accessible in your system PATH.
• The Chrome extension is loaded locally and is NOT YET PUBLISHED on the Web Store.
• In backend/server.py, the Whisper model "small" is loaded by default. Users can change it to other models based on their system’s processing power:

Model	Approx. RAM Required	Recommended Use Case
tiny	~1 GB	Low-resource machines, faster processing
base	~2 GB	Lightweight, reasonable accuracy
small	~4 GB	Default, good balance of speed and accuracy
medium	~8 GB	Higher accuracy, slower processing
large	~16+ GB	Maximum accuracy, requires powerful CPU/GPU

• Users can adjust the model in server.py according to their available RAM and processing power.

import whisper

# Change the model here:
# Options: "tiny", "base", "small", "medium", "large"
model = whisper.load_model("small")