I built Chief of Staff (CoS) so I could listen to high quality text-to-speech (TTS) audio readings of articles while my hands and eyes were busy doing other things.
The first demo took a few hours, but what you see today represents over two years work. I've included tech stack notes and a simple system architecture diagram [1] if you're curious.
So, why use it?
CoS is pretty fast. Articles begin synthesis the moment you add them are often ready to listen before you can finish humming Happy Birthday once or twice. If an article has already been synthesized by another user, the audio is made available to you instantly.
The reader voice sounds good. You can hear a short article on the public landing page. New accounts are free and the onboarding wizard helps you add a longer article about Paul Simon’s 1986 album Graceland from Rolling Stone. (An interesting listen!)
The CoS interface looks and feels good. The audio player remembers playback position. So, you can start listening on your laptop, close the tab then pop the site up on your mobile device and pick up right from where you left off.
I started working on this product just before our first baby was born in June 2020. I frequently would come across articles on HN that I wanted to read but could not create screen time for.
Initially, I used it for long walks with my leash-trained cat. Later, it would come in handy for many late nights with an infant (which also necessitated a decent dark mode!)
I wanted something to provide the save-it-for-later feel of bookmarking while making consumption more flexible, (Hands and eyes free). Finally, I wanted it to have a “check it off when it’s done” feel, so I built archive and delete buttons so you can keep your unread content area clean.
This workflow may appeal to HN readers who live with ADHD. There was a post a few months ago where a commenter described roughly what I’ve built as a coping mechanism. [2]
Free accounts can synth a few articles depending on length.
I’ve set an introductory price for pro accounts that is cheap in comparison to other TTS products. You can also enjoy a week of free trial of the pro account. Subscription cancellation behavior mirrors that of the App Store (cancel any time and enjoy sub through expiration).
Chief of Staff’s technology stack is: - Python / Django backend with DRF (Django rest framework) API - Vanilla JS and typescript frontend - Postgres database
The product leverages hosted cloud service: - Mailgun (transactional email) - Posthog (analytics) - Stripe (payment processing) - Unsplash (to find ~related images when article lacks one such as PG's articles) - AWS (Polly and SQS for TTS, S3 for images and media)
In devops land: - Github actions (CI/CD) - Frontend bundling via esbuild
I’ve followed TTS and generative AI developments with keen interest since beginning this work and am hopeful to flesh out its vision as a super-Siri, personal alpha geek, or ahem, Chief of Staff, that keeps you briefed on what matters most.
But for now it is a humble, (hopefully) well-designed article to speech tool I built first for myself and wholly in my spare time.
I hope you like it and appreciate any questions, suggestions or other feedback.
rob - [email protected]
