HackerNews Podcaster

A TypeScript application that converts top Hacker News stories into podcast audio using AI. Works with any OpenAI-compatible API.

About

Inspired by Hacker News Recap by Wondercraft.ai, this tool fetches top stories from Hacker News, summarizes them using AI language models, generates a conversational podcast script, and converts it to audio.

The project works with any OpenAI-compatible API provider. You can use OpenAI's models, xAI's Grok models, or any other compatible service. Simply configure the base URL and model names - no code changes needed.

Example Output

• April 14, 2025 Podcast (Transcript)
• March 25, 2025 Podcast (Transcript)
• November 7, 2023 Podcast (Transcript)

Features

• Fetches top stories from Hacker News
• Generates summaries and talking points using AI
• Creates conversational podcast scripts
• Iteratively refines scripts for clarity
• Converts scripts to MP3 audio
• Saves transcripts and audio to the output directory

Setup

1. Clone the repository

git clone https://github.com/amscotti/hn-podcaster.git
cd hn-podcaster

2. Install Deno

Install Deno if you haven't already:

curl -fsSL https://deno.land/x/install/install.sh | sh

Or use mise which will automatically use the version specified in mise.toml:

mise install

3. Configure API Keys

Create a .env file in the project root. The example below uses xAI's Grok models for text generation and OpenAI for text-to-speech:

# General AI (for summaries and script generation)
GENERAL_AI_API_KEY=your_api_key
GENERAL_AI_BASE_URL=https://api.x.ai/v1
GENERAL_AI_MODEL_SUMMARY=grok-4-fast-non-reasoning
GENERAL_AI_MODEL_MAIN=grok-4-fast

# Text-to-Speech
OPENAI_API_KEY=your_openai_key
OPENAI_TTS_BASE_URL=https://api.openai.com/v1
OPENAI_MODEL_TTS=gpt-4o-mini-tts

Using other providers: Any OpenAI-compatible API works. Just set the GENERAL_AI_BASE_URL to your provider's endpoint and use their model names. For example, to use OpenAI for everything, set GENERAL_AI_BASE_URL=https://api.openai.com/v1 and use model names like gpt-4o or gpt-4o-mini.

4. Run

Generate a podcast:

deno task start

Output files are saved to the output/ directory.

Development

Type checking, linting, and formatting:

deno task check

Run tests:

deno task test

Project Structure

• src/ - Core application code
  • services/ - Service layer with interfaces and implementations
  • workflows/ - High-level workflow orchestration
  • shared/ - Shared utilities (errors, logging)
• app.ts - Main entry point
• output/ - Generated podcasts and transcripts
• example/ - Example outputs

Architecture

The project uses a service-based architecture with dependency injection. AI providers are abstracted behind interfaces, so you can use any OpenAI-compatible API (OpenAI, xAI, locally hosted models, etc.) by just changing the configuration - no code changes required.

Core components:

• HackerNews.ts - Fetches stories from the Hacker News API
• PodcastCreator.ts - Orchestrates summarization and script generation
• PodcastRecorder.ts - Converts scripts to audio
• workflows/ - Coordinates the full pipeline with error handling

Requirements

• Deno 2.5+
• API key for any OpenAI-compatible provider (OpenAI, xAI, etc.)