Read To Me

Under ConstructionBuilt with AI

Long articles → audio for your commute. Built because I wanted podcast-quality narration of web articles.

The Problem

I was commuting 45 minutes daily and wanted to 'read' long articles without looking at my phone. Existing TTS apps had terrible voices or clunky UIs. I just wanted to paste a URL and listen.

What I Built

Drop in any URL, article, or PDF → get clean audio narration with a great British voice. Built in one weekend (~8 hours).

Real Usage

12 friends using it for commutes. One friend listens to 3-5 articles per day at 1.5x speed. Personally use it every morning for newsletter catch-up.

Tech Stack

Frontend

Backend

AI/ML

LLMTTSChunkingEval harness

What I Learned

Voice Quality Matters

TTS voice quality matters more than speed-to-audio conversion time

Chapter Breaks Needed

People want chapter breaks for long content (15+ min)

Text Cleanup is Key

Text cleanup/normalization is 60% of the work

Wait Time Tolerance

Users will tolerate 2-3 second wait for good narration quality

Where It Could Go

Next Features

✓AI summarization before playback
✓Multiple voice options with quality evals
✓Mobile app with offline playback
✓Playlist management and auto-queue from RSS feeds
✓Integration with read-later apps (Pocket, Instapaper)

Bigger Vision

A comprehensive audio-first content platform that transforms any written content into engaging, podcast-quality experiences with AI-powered enhancements.