Spaces:
Running
Running
| title: URL To Audio Summary | |
| emoji: 🐠 | |
| colorFrom: gray | |
| colorTo: pink | |
| sdk: gradio | |
| sdk_version: 5.34.2 | |
| app_file: app.py | |
| pinned: false | |
| license: mit | |
| short_description: URL to Audio Summary Agent | |
| # 🗣️ URL to Audio Summary Agent (LangChain + Hugging Face) | |
| This project is an advanced LangChain-powered agent that takes any article/blog URL, summarizes it using a CPU-friendly language model, and generates a spoken audio summary. Designed to run entirely on **CPU**, it's perfect for deploying on **Hugging Face Spaces**. | |
| --- | |
| ## 🚀 Features | |
| - 🌐 Fetches and parses web content from any URL | |
| - 🧠 Summarizes long articles using `flan-t5-base` | |
| - 🔊 Converts summaries into speech using `espnet/kan-bayashi_ljspeech_vits` | |
| - 🖥️ CPU-only — no GPU or API keys required | |
| - 🧱 Built using LangChain, Gradio, Transformers | |
| --- | |
| ## 🛠 Tech Stack | |
| - **LangChain**: Document loader, prompt chaining | |
| - **Transformers**: `flan-t5-base` for summarization | |
| - **ESPnet/VITS**: Natural-sounding voice TTS model | |
| - **Gradio**: Easy interface for Hugging Face Spaces | |
| --- | |
| ## 🧩 Usage | |
| 1. Paste any blog or article URL (e.g. from Medium, BBC, etc.) | |
| 2. The agent will: | |
| - Load and extract article content | |
| - Summarize the article intelligently | |
| - Convert the summary into audio | |
| 3. You get both: | |
| - 📄 Text Summary | |
| - 🔉 Downloadable Audio Summary (WAV) | |
| --- | |
| ## 📦 Installation | |
| To run locally or on Spaces: | |
| ### `requirements.txt` | |
| ```txt | |
| gradio | |
| langchain | |
| transformers | |
| torch | |
| sentencepiece | |
| beautifulsoup4 | |
| ``` | |
| --- | |
| ## 💡 SEO Tags | |
| `AI article summarizer`, `Text-to-Speech summarizer`, `LangChain agent`, `Hugging Face CPU summarizer`, `URL to audio`, `audio content AI`, `gradio langchain tts` | |
| --- | |
| ## 🧠 Example Models Used | |
| - `google/flan-t5-base` (Summarization) | |
| - `espnet/kan-bayashi_ljspeech_vits` (TTS) | |
| --- | |
| ## 📍 Ideal For | |
| - Podcasters turning news into voice | |
| - Accessibility tools | |
| - EdTech content summarization | |
| - Personal AI readers | |
| --- | |
| ## 🪪 License | |
| MIT — fork, remix, and deploy freely. | |
| --- | |
| ## 🙌 Credits | |
| Built using [LangChain](https://www.langchain.com/) and [Hugging Face Transformers](https://huggingface.co/transformers/). | |