News Summarization & Hindi TTS Application

Shaik Nawaz

Completed work

Data Analyst

BeautifulSoup

NLTK

Python

Electronics

News Summarization & Hindi TTS - Documentation

Project Overview

This project is designed to extract news articles related to a given company, summarize them, analyze their sentiment, translate the summary into Hindi, and generate a Hindi text-to-speech (TTS) output. The application is built using Streamlit for the frontend, with various NLP techniques implemented in the backend.

Features

Fetches latest news articles using NewsAPI

Summarizes the extracted news content

Performs sentiment analysis on summarized text

Extracts key topics from the summary

Translates the summary into Hindi

Converts the Hindi summary to speech using Google TTS

Provides an interactive UI using Streamlit

Tech Stack

Frontend: Streamlit

Backend: Python (Flask APIs, NLP models)

APIs: NewsAPI for fetching news articles

Libraries Used:

requests for API calls

nltk for sentiment analysis

transformers for text summarization

keybert for key topic extraction

googletrans for language translation

gTTS for text-to-speech

beautifulsoup4 for web scraping (if required later)

Installation & Setup

1. Clone the Repository

git clone https://github.com/your-repository-link.git
cd news-summarization-hindi-tts

2. Install Dependencies

pip install -r requirements.txt

3. Run the Application

streamlit run app.py

Code Structure

news-summarization-hindi-tts/
│── api.py                   # Fetches news articles using NewsAPI
│── utils.py                 # Contains summarization, sentiment analysis, translation, key topics extraction, and TTS functions
│── app.py                   # Streamlit frontend application
│── requirements.txt         # List of required dependencies
│── README.md                # Documentation

Model Details

Summarization Model

Uses a pre-trained transformer model from Hugging Face (e.g., facebook/bart-large-cnn).

Extracts key points from the news article.

Sentiment Analysis

Uses NLTK's VADER sentiment analysis model.

Categorizes text as Positive, Neutral, or Negative.

Key Topics Extraction

Uses KeyBERT to extract the most relevant topics from the summary.

Hindi Translation

Uses Google Translate API (googletrans) to convert English summaries to Hindi.

Text-to-Speech (TTS)

Uses gTTS (Google Text-to-Speech) to generate an MP3 audio file of the Hindi summary.

API Development

The project includes APIs for fetching news, summarization, sentiment analysis, key topics extraction, translation, and text-to-speech conversion.

API Endpoints

1. Fetch News

Endpoint: fetch_news(company_name)

Description: Fetches the latest 5 news articles about the given company.

Response Format:

[
  {
    "title": "Article Title",
    "key_topics": ["Topic1", "Topic2"],
    "summary": "Short summary of the article.",
    "sentiment": "Positive",
    "sentiment_score": 0.1779,
    "hindi_summary": "हिन्दी में सारांश"
  }
]

2. Summarization

Function: summarize_text(text)

Description: Summarizes the input text using a Hugging Face transformer model.

Returns: Shortened summary of the input text.

3. Sentiment Analysis

Function: analyze_sentiment(text)

Description: Uses NLTK's VADER sentiment analysis to classify text as Positive, Neutral, or Negative.

Returns: Sentiment category and sentiment score.

4. Key Topics Extraction

Function: extract_key_topics(text, num_keywords=3)

Description: Extracts the most important topics from the given text using KeyBERT.

Returns: A list of key topics.

5. Hindi Translation

Function: translate_to_hindi(text)

Description: Translates the given text from English to Hindi using Google Translate API.

Returns: Hindi-translated text.

6. Text-to-Speech (TTS)

Function: text_to_speech_google(text, filename="news_summary.mp3")

Description: Converts Hindi text into a speech audio file.

Returns: Filename of the generated MP3 file.

API Usage

The APIs can be tested using Postman by making requests to the backend service.

Example Postman Request for fetching news:

GET http://localhost:5000/fetch_news?company_name=amazon

Assumptions & Limitations

Assumptions

NewsAPI provides relevant articles based on the given company name.

The summarization model correctly captures the essence of the news article.

Google Translate provides an accurate Hindi translation.

TTS output is clear and understandable.

Limitations

NewsAPI has rate limits, so too many requests may fail.

Sentiment analysis might not be 100% accurate for complex news content.

Google Translate may not always provide perfect translations.

gTTS might have pronunciation issues for certain Hindi words.

Deployment Plan

Local Testing: Ensure all functionalities work as expected.

Containerization (Optional): Use Docker to package the application.

Deployment on Hugging Face Spaces:

Push code to GitHub

Deploy using Hugging Face Spaces

Test and verify

Future Improvements

Support for multiple languages.

Improved summarization with fine-tuned models.

Enhanced sentiment analysis with deep learning models.

Integration with real-time news sources.

This documentation will be continuously updated as new features are added.

Like this project

Completed work

Posted Sep 7, 2025

Developed a news summarization and Hindi TTS application using Streamlit and NLP techniques.

Likes

Views