I built a tool to convert YouTube videos into podcasts

Body

Problem: I kept queuing YouTube tutorials and talks but never watching them. Video demands attention in a way that audio doesn’t.

Solution: VoxTube extracts transcripts from YouTube videos and converts them to audio using high-quality TTS.

Now I "watch" YouTube during my commute, while cooking, and during workouts.

Technical details:

• Built with Bun + Hono (~300 lines)
• Uses Kokoro TTS (runs locally via Docker)
• Caches generated audio
• No cloud dependencies

What I learned:

• Bun’s file APIs are really nice for streaming audio
• Modern TTS (Kokoro) sounds surprisingly natural
• Most YouTube videos have transcripts available

Stats:

• 2 weeks to MVP
• ~300 lines of code
• 0 monthly costs (runs locally)

GitHub: https://github.com/shawn-dsz/voxtube

Happy to answer questions about the build!