Voice RSS Summary

Voice RSS Summary automatically collects RSS feeds, summarizes new articles using OpenAI, and publishes them as a podcast with speech synthesized by VOICEVOX. The project uses the Bun runtime with a React (Vite) front end and an SQLite database.

Features

Fetch multiple RSS feeds listed in feed_urls.txt.
Generate podcast scripts with the OpenAI API.
Convert scripts to audio using VOICEVOX and FFmpeg.
Serve a web interface to browse feeds and play episodes.
Provide a podcast RSS feed at /podcast.xml.

Prerequisites

Bun 1.x (if running locally)
VOICEVOX engine
An OpenAI API key

Installation

Install dependencies:
```
bun install
```
Build the front end:
```
bun run build:frontend
```

Create a .env file and set the following variables:

OPENAI_API_KEY=your-api-key
OPENAI_API_ENDPOINT=https://api.openai.com/v1
OPENAI_MODEL_NAME=gpt-4o-mini
VOICEVOX_HOST=http://localhost:50021
VOICEVOX_STYLE_ID=0
PODCAST_TITLE=Voice RSS Summary
PODCAST_LINK=https://your-domain.com/podcast
PODCAST_DESCRIPTION=Generated podcast feed
PODCAST_LANGUAGE=ja
PODCAST_AUTHOR=Admin
PODCAST_CATEGORIES=Technology
PODCAST_TTL=60
PODCAST_BASE_URL=https://your-domain.com
FEED_URLS_FILE=feed_urls.txt

Prepare feed_urls.txt with one RSS feed URL per line.
Start the server:
```
bun run server.ts
```

The initial batch process runs on start and then every 24 hours.

Docker

A Dockerfile and helper scripts are provided. To run in a container:

./build-docker-image.sh
./run-docker.sh

These scripts mount feed_urls.txt, .env, public/, and data/ for persistence.

Directories

data/ – SQLite database (podcast.db)
public/podcast_audio/ – generated MP3 files
frontend/ – Vite React application (served on build)

Access the web UI and podcast feed on http://localhost:3000/ after the server starts.

License

This project is licensed under the Apache 2.0 License.

2.1 KiB Raw Blame History Unescape Escape