crsr

prior version used modal to host the sam audio model (https://github.com/sambarrowclough/clearaudio/pull/2). fal made things simpler & faster

alternative open source version here: https://github.com/sambarrowclough/clearaudio

I built ClearAudio to make prompt-based audio cleanup practical. Upload audio, describe what you want ("keep the voice, remove the crowd"), and it separates it out. I stripped it down to the essentials with a minimal 1-bit UI—no extra features, just the core separation.

Uses Meta's AudioSeg model (https://github.com/facebookresearch/audioseg) under the hood. Also used this as a chance to learn Modal for GPU inference.

Code is open source: https://github.com/sambarrowclough/clearaudio

Happy to answer questions about the stack or take feedback on the UX!

Browserless have browsers as a service, and a dedicated pdf endpoint[0] you can call. Had really good experience with this.

[0] http://docs.browserless.io/rest-apis/pdf-api

Nice! I built something similar yesterday. But decided to use elysia instead. They have a tRPC like client with full TS support called eden.

- eden: https://elysiajs.com/eden/overview.html

I recently created a product that has picked up some traction. The name of the tool is a character from a famous comic book. The characters name is trademarked. I'm from the UK. My tool is an analytics tool. Does this run the risk of me getting sued by the comic book company?

Cool idea, well executed. How do you compare with repl.it?

What would your response be?

Just a one man band or a couple of people who are bootstrapping an idea, living reasonably comfortable from the money it makes

Used supabase on a few side projects. Overall had a great experience with it. Would recommend.

a few to get you started: - https://github.com/vercel/nextjs-subscription-payments - https://bedrock.mxstbr.com/

Bill Atkinson reminds me of Gilfoyle from Silicon Valley

This is brilliant. I was wondering how this tool worked.

Glad you like!

Thank you!

just added!

Yes your right, missed that. Just updated the repo: https://github.com/sambarrowclough/pricing-page-24h/blob/mas...

I think this idea only really speaks to the Roam Research crowd, which is quite esoteric. Non the less, it is cool project.

This is a problem that Google tackled themselves, but it had too many problems[0]. Recent attempts use Chrome Dev Tools[1] to cache the results. So this may alleviate problems Google had. Memex had attempted this, but deprecated it:

> We realized although its a valuable feature to search your browsing history, its not solving a super frequent and painful problem for users

...

> We spent so much time on building the search and with it also 30-50% more time on every new feature because of interdependencies with the amount of data produced. End2End encrypted sync, backup, search performance, search filters, all were directly or indirectly necessary to work much better than we can afford. We're just 2.5 devs.

Another developer attempted this, too. But didn't have many users:

> I'll also point out that I did collect usage stats for a time, and they were horrific. At my peak I had ~5000 installs and out of those 5k something like 3-5 searches/day was the norm.[2]

One user does point out the reason why this may not have took off:

> about once every two months, I am looking for something that I swear I came across on the Internet at some point. However, the rest of the time, I'm able to re-find it just by doing another search, whether on search-engine-of-choice, or a search box on particular-website (e.g. socnet, stackoverflow, reddit, github, hacker news...).[3]

[0] https://news.ycombinator.com/item?id=17745931

[1] https://github.com/c9fe/22120

[2] https://news.ycombinator.com/item?id=13427464

[3] https://news.ycombinator.com/item?id=17744785

> Most people use vim in two stupid dimensions. But not me. I use it in three

Brilliant.