We built this because
we had a reading list too.
Three people. A small studio on Mass Ave in Cambridge. A dog named Sigma who sleeps under the desk.
We started in the summer of 2023 because we had the same problem everyone else has: a reading list that kept growing, and time that kept shrinking. Open tabs, pocket articles, bookmarks folder nobody touched. Promises to a future self that never showed up.
So we built a thing to read the pieces out loud. Not in the voice of a customer service bot. In voices you'd actually want to hear. Someone who sounds like they've read the piece before you. Someone who pauses where you'd pause.
The first version was ugly. One voice, one article at a time, a button that said "make mp3". We gave it to fifty friends. Journalists, runners, a couple of blind readers who wrote back and told us what to fix. Most of what's in the product now came from those first emails.
The team is still small. Three of us full-time. A couple of voice engineers on contract. No investors yet. We ran our own savings down for the first year and put it back from revenue. We like it that way.
Hear The Web is not a startup pretending to be a studio. We ship when it's ready. We read the articles ourselves before training a voice on them. We answer our own support email. When you sign up, one of the three of us will probably see it.
Three years, told short.
Three humans.
One studio dog.
Delphine P.
voice design · co-founder
Used to direct audiobook narrators in Paris. Handles the voice library and the phrasing engine. Wrote most of our style guide.
Mats V.
engineering · co-founder
Speech ML for ten years. Previously at a research lab in Stockholm. Owns the render pipeline, the model training, and the fact that long articles do not stutter.
Rafi K.
everything else · co-founder
Design, apps, billing, the website, the support inbox. If you email us you probably get Rafi. If you get a cold reply from a form, yell at Rafi.
A few things we hold to.
- Voices before features. One good voice beats ten mediocre ones. We shipped six before we shipped any feature.
- No training on your content. Articles you render are never used to train anything. They exist long enough to become audio. Then they're gone.
- Accessibility is not a feature. Low-vision readers were our first testers. Anything that doesn't work for them doesn't ship.
- Free stays free. Three articles a day. We don't trial-bait. If you never pay us, you still get to listen.
- We answer our own email. Actually.
===============================================================================