🌐 EqualVoice

EqualVoice is an AI-powered text analysis platform designed to combat gender bias in communication. It acts like Grammarly for DEI — providing real-time feedback on how inclusive and gender-fair your language is. Whether you're a journalist, product designer, activist, or educator, EqualVoice helps you write responsibly with fairness, awareness, and inclusivity.

---

✨ Why EqualVoice?

Stakeholder	How EqualVoice Helps
Writers & Bloggers	Instantly flag gendered terms, receive bias scores, and get rewrite suggestions.
Students & Educators	Learn gender-aware writing and track inclusivity improvements over time.
NGOs / Activists	Analyze speeches, news, or policy docs for systemic bias.
Product & UX Teams	Audit app content and marketing copy before shipping.
Journalists & Media Houses	Quantitatively measure gender representation in reporting.

---

🛠 Tech Stack

🎨 Frontend

---

🧠 Backend / ML Microservice

---

🚀 Deployment

---

📸 Screenshots

---

🔍 Core Features

1. Real-Time Text Analysis

Paste or upload any text → receive Bias Score, Feminist Theme Score, and highlighted phrases instantly.

2. Rewrite Suggestions

Get smart replacements for stereotyped phrases like:

• “bossy” → “confident leader”
• “emotional” → “expressive”

3. Visual Sentiment Insights (coming soon)

Sentiment by gender group, radar charts for bias metrics, and inclusivity tracking.

4. Developer-Friendly API

POST text to /analyze and get JSON with metrics, suggestions, and highlights.

---

🧠 How It Works

1. Tokenization with spaCy
2. Gender Entity Matching using keyword lists + gender-guesser
3. Sentiment Analysis via VADER on gendered sentences
4. Bias Score Calculation based on gender balance + sentiment
5. Feminist Theme Detection using PhraseMatcher
6. Rule-Based Rewrite Suggestions (e.g. “nagging” → “persistent”)

---

📂 Folder Structure

equalvoice/
├── frontend/          # Vite + React + Tailwind
├── ml-service/        # FastAPI + NLP logic (spaCy, VADER, gender-guesser)

---

🚀 Getting Started

1. Clone the Repo

git clone https://github.com/SanyaShresta25/equalvoice.git
cd equalvoice

2. Run the ML Backend (FastAPI)

cd ml-service
pip install -r requirements.txt
python -m spacy download en_core_web_sm
uvicorn main:app --reload
# → http://localhost:8000/docs

3. Run the Frontend (React)

cd frontend
npm install
echo "VITE_API_URL=http://localhost:8000" > .env
npm run dev
# → http://localhost:5173

---

📤 Example API Call

POST /analyze
Content-Type: application/json

{
  "text": "She is a confident leader, but some say she is bossy."
}

Response:

{
  "biasScore": 76.5,
  "feministThemeScore": 40.0,
  "genderMentions": { "male": 2, "female": 6, "neutral": 53 },
  "highlightedPhrases": [
    { "phrase": "bossy", "type": "stereotype", "startIndex": 47, "endIndex": 52 }
  ],
  "suggestions": [
    { "original": "bossy", "suggestion": "confident leader" }
  ]
}

---

🤝 Contributing

• Fork the repo
• Create your branch: git checkout -b feat/feature-name
• Commit your changes
• Open a PR describing what you did and why

---

📜 License

MIT License © 2025 EqualVoice — crafted by Sanya Shresta

---

“Inclusive language is not a privilege, it’s a practice.”

---