this website reads anything to you

Added: 2026-05-06

[00:00:00]Day two, bankrupting million-dollar companies by building shitty website alternatives.

[00:00:06]And today's target is Narration text-to-speech. A common player is ElevenLabs. And basically how it works is you type in your text or you upload a text file, the software synthesizes what the text should sound like as speech, and then these companies slam you with a cheeky paywall before you get any positive value. Slight tangent, my favorite subscriptions are probably World of Warcraft, a gym membership, and Claude Code. And some honorary would be Spotify, uh the domains that I rent, and also Canva. But my goodness, when I go and check my bank statements and see all of these subscriptions, basically holes in my wallet, the hurt is definitely greater than the sum of its parts, that's for sure. And because Claude Code is one of my subscriptions, I literally have access to the best coder on the face of the planet. The mission is simple, we're going to build a shitty free alternative that does a couple of voices really well. It's going to be totally free to use, no cost, no ads, no sign-up, or absolutely anything. And technically, that has infinite cost-benefit analysis over a paid service, even if it's not quite as good, as long as we can get 80% of the way there, then I think it's an absolutely brilliant product. Now, for this project to be feasible, there's three criteria that basically we need to meet. One, I need to find an email model for Narration. Two, it needs to have a number of voices available cuz it's pretty boring if it's just, you know, standard John voice every single time.

[00:01:28]Ideally, it would have about six to eight different voices so that people actually feel like there's some substance to the service. And number three, we have to find a way, engineer a way to run this whole system in the browser, so that I don't have to have any server-side code because that's what would incur me cost, which would incur you cost. So, to keep it totally free, everything has to run on the client, on the user's computer. Turns out it was in fact possible. After a fairly lengthy discourse with Claude Code, we found a way. Initially, the biggest problem, since it's a shitty website, was not actually developing the website. The website itself was very rudimentary. The problem was finding these machine learning models.

[00:02:07]And because a lot of the machine learning models have outdated knowledge, it was pointing me to all these links, all these downloadables, all these machine learning models that either didn't exist or had been privatized. So, the use of them was now blocked behind a paywall, a sign-in, a login, something like that, which once again just completely defeats the purpose of a free-to-use, no registration needed service. So, after some back and forth, eventually Claude and I had this genius idea to get it to do some internet research. And so, Claude ran off. It was like, "James, I'm going to go to the internet and look for some more search results to see if I can find where's the free models have gone."

[00:02:47]Surely Claude sounds like a Germanic name. I feel like that's appropriate.

[00:02:52]And it went off, and eventually it found exactly what we needed, models that are free to use we can pull from the internet. And what we can do is when a user accesses the website, we can download the model to their browser just once, which takes a tiny bit of time, and then we can cache it so that in future the download is already finished, and they can get straight to the speech synthesis.

[00:03:12]Now, this process takes a little bit of time because since it's running on a user's computer, it's also dependent on the specs of the computer. But once again, 80% of the full product for free is still infinite cost-benefit analysis.

[00:03:26]So, the question becomes, how did it all pan out? Well, >> what I thought I would do is use a previous shitty website that we made for transcription services to transcribe the audio from this video, and then upload it to Big Daddy Read to Me, and have this whole transcript synthesized so that I could overlay it instead of my actual voice once again as a proof of product.

[00:03:47]>> all you have to do is never pay for dictation ever again. Also, let me know in the comments down below what you think of Big Daddy Read to Me. And finally, if you have any ideas for shitty websites or you want any other companies to be completely obliterated with brilliant shitty free alternatives, let me know in the comments down below.

[00:04:04]Catch you guys later. Peace.