The Problem: Like many of you, I subscribe to way too many long-form newsletters (Substack, technical digests, etc.), but I never find the time to actually sit down and read them. I tried using standard screen readers, but listening to them read URLs, "view in browser" links, and footer disclaimers was unbearable.
The Solution: I built a tool that creates a private podcast feed from your emails.
1. Ingestion: You get a unique email address. We use Cloudflare Workers to catch the inbound mail instantly.
2. AI Cleaning: We don't just read the raw text. We run a 3-stage pipeline (Cheerio -> Mozilla Readability -> GPT-4o-mini) to strip the "email junk," remove tracking pixels, and fix the formatting.
3. Hybrid TTS: We prioritize a self-hosted instance of Kokoro TTS (running on a Reserved VM) for high-quality, low-latency audio, with a fallback to Google Cloud Neural2 for resilience.
4. RSS: It generates a standard podcast feed you can add directly to Apple Podcasts or Overcast.
How it works:
Forward an email to providedemail@forwardtoaudio.com (sign up to get your own permanent address). It gets processed and appears in your feed minutes later.
Tech Stack:
* Backend: Express/TypeScript with Drizzle ORM & Postgres.
* Audio Engine: Self-hosted Kokoro (Primary) + Google Neural2 (Failover circuit breaker).
* Parsing: Mozilla Readability for extraction, LLMs for polish/SSML insertion.
It’s free to try (no credit card). I’d love to know if the parsing handles your weirdest newsletters correctly—email HTML is a beast!
Hi HN, I’m Bryan founder of ForwardToAudio.
The Problem: Like many of you, I subscribe to way too many long-form newsletters (Substack, technical digests, etc.), but I never find the time to actually sit down and read them. I tried using standard screen readers, but listening to them read URLs, "view in browser" links, and footer disclaimers was unbearable.
The Solution: I built a tool that creates a private podcast feed from your emails. 1. Ingestion: You get a unique email address. We use Cloudflare Workers to catch the inbound mail instantly. 2. AI Cleaning: We don't just read the raw text. We run a 3-stage pipeline (Cheerio -> Mozilla Readability -> GPT-4o-mini) to strip the "email junk," remove tracking pixels, and fix the formatting. 3. Hybrid TTS: We prioritize a self-hosted instance of Kokoro TTS (running on a Reserved VM) for high-quality, low-latency audio, with a fallback to Google Cloud Neural2 for resilience. 4. RSS: It generates a standard podcast feed you can add directly to Apple Podcasts or Overcast.
How it works: Forward an email to providedemail@forwardtoaudio.com (sign up to get your own permanent address). It gets processed and appears in your feed minutes later.
Tech Stack: * Backend: Express/TypeScript with Drizzle ORM & Postgres. * Audio Engine: Self-hosted Kokoro (Primary) + Google Neural2 (Failover circuit breaker). * Parsing: Mozilla Readability for extraction, LLMs for polish/SSML insertion.
It’s free to try (no credit card). I’d love to know if the parsing handles your weirdest newsletters correctly—email HTML is a beast!