Hacker News new | ask | show | jobs
Show HN: Descript wanted $24/mo, I built an open-source alternative in a weekend (github.com)
39 points by wassimgr 2 days ago
Descript costs $24/mo, so I built this over a single weekend with Fable!

Introducing Rescript: edit videos by simply editing the transcript text. Drop in a video and it is transcribed locally with per-word timestamps and speaker labels. Delete words in the transcript and the corresponding clip is cut from the video.

Runs fully in the browser: Local, free, offline and open source.

→ Github https://github.com/wassgha/rescript

→ App https://wassgha.github.io/rescript

15 comments

Have you tried using this to actually edit anything? The local transcription is trash (slow, tons of repeated words, no speaker diarization, no option to upload a transcript or use ElevenLabs). I made a few edits to a sample episode I have, clicked on the upper left for a menu, it went to the front page and...I guess it lost the transcription and the edits with no way to get it back. Cool idea but it seems unusable at the moment, I'd say you have a lot more weekends of work to have something suitable for either Show HN or Product Hunt.
Just pushed a PR that fixes hallucinations and lets it handle longer form video + the option to select a larger (~600mb) model that has much better transcription accuracy.

Please give it another try! https://github.com/wassgha/rescript/issues/13#issuecomment-5...

Also just added the ability to save/restore projects! :)
Posts like these now feel the equivalent of “I didn’t want to order Indian takeaway so I hired a chef to make me butter chicken”
But this guy will probably get hired because of this.
He works at Tesla already.
yes, and the chef was a robot, so less of an issue with them in your house, and the robot only cost 50 cents.

(seeing the comment that it took 4 hours or so of AI time)

that's exactly what I see.

Yep, a very subsidised chef
yes. and I tell everyone that I expect that to change, so don't get used to the current status quo. We won't be able to build with top notch semi-unllimited robots for 20-$30 per month after the IPOs I am sure..

However we are seeing models come out like Gema and the distilation accused that are likely capable of building a browser based Descript, or Quickbooks (as most people would use them) and many other saas type things.. just running on a walmart PC for free at this point and going forward right?

So the future is now, the platform and url may change, but it looks like most software and art will be reproducible for 50 cents or a bit more one way or another. Maybe not as polished, maybe missing a few things that elitists point to.

That's not going to replace the software subscriptions for the fortune 500. Yet it's already replaced several saas monthly / yearly payments for me.

It took 4 hours and 20 commits :)
That would make a good butter chicken
This is by far the least interesting/worst kind of project emerging lately.

"I didn't want to pay trivial money for a product, so I paid nontrivial money to have an AI make a version that's worse in every way that matters"

Who cares?

But they own this version instead of renting it. And it’s been released for free to the world.

That’s a way that the “worse” version is better to some people.

Also, assuming this person used $200/month subscription to some robot, and they used 5 days of of their quota, the money comes out about even after the first month or two, no?

Please ignore the attention-grabbing title, the novelty here is that it runs fully in the browser with local/offline models that showcase some of the new browser technologies. This is not meant to be as replacement / fully-featured product, but rather a tech demo and an argument for what "personal software" can look like in the future.
Follow https://x.com/wassgha for updates/new releases.

Currently on the roadmap:

- Saving/re-opening projects – done

- Importing transcripts (SRT / VTT / JSON) – done

- Support for audio/podcasts – done

- Importing plain text transcripts (auto sync)

- Manually editing word boundaries

- Splitting / reordering of scenes

- Desktop app

- Support for Apple SpeechAnalyzer to remove the need to download a speech model

I will never understand that thinking. You didn’t like that a service you need asks to be paid, so you paid an LLM to create a free competitor that you now advertise online. Do you not see the race to the bottom you’re contributing to? $24 doesn’t sound outrageous at all.

Such a weird way to frame your project

It's not $24 - it's $24 / month.

And last time I used it, when I went to export in high resolution (don't recall if it was 1080 or 4k) - it refused to export in the resolution that I uploaded unless I paid for an upgrade.

I can't stand monthly rent forever, yet a time many moons ago, descript's method was novel enough to give it a whirl.

It's no longer novel, just like many other projects.

The title is largely for attention-grabbing + relating the project to a product people know already; obviously this is not a replacement, the goal is to experiment with local/offline models and test the limits of the browser and showcase some of the new technologies like SharedArrayBuffer. I work at the limit of javascript performance and making projects like these is how I teach myself new browser APIs.
If it’s something you’ve done yourself, and not Claude generated, then kudos, that’s definitely the good way to learn
It's was always a case to describe your idea in context already known to users. Examples Uber for X, or eBay for Y. I think it's a great way to learn to build alternatives for payed services. Look at cal.com
4 hours and 20 commits with Cursor is wild for getting something like this off the ground. How much of that time was spent fighting with prompt context vs actually structuring the browser media APIs?
How does this perform on longer videos, especially on lower-end machines? Is the transcription and video processing entirely local in the browser?
Whilst I love opensource; hitting with $24/mo saving when you invest 100x that much in your time is probably not the way to sell it.
Where it pays off is not when you reach parity with the built product compared to the vendored product - it's when you can continue to extend or customize the capabilities when the original vendor can not or will not support the features you can now build
I promise you something like this will never monetarily "pay off."
I don't think they mean "monetarily". But once you have some reasonable parity you can implement your own features/workflow as you want them and not be reliant on Descript.
The context is i.r.t. the monthly cost and the cost of one's time, which has a monetary value, as well as the cost of tokens. So I think it's rightful to assume the topic is of a monetary pay off.
From the commits, it looks like it only took OP about 4 hours with Cursor.
^ this :)
Interesting, i have the same goal except it's for daily productivity tool github.com/slaveofcode/goodwebtools
The name has a collision with a programming language related to ocaml - ReScript, which evolved from ReasonML
Looks like Im the only person who actually tried it? Very underbaked, bordering on useless - single word deletions dont even register. Multi word deletions have the first half/syllable of the first word heard. When you remove anything, there is an abrupt skip in the audio making the output weird at best and unusable for any real production use case, even for personal needs.
Hey @dcreater, I’ve just pushed a big update that adds a lot of manual controls over the start/end of words as well as the ability to manually split clips and insert skips. I also deployed a larger, more accurate speech model (whisper small); you can use the model selector to try it out! https://github.com/wassgha/rescript/
So you collect now 22,99 USD / month? :-D
one peanut / month is the current price
Can you enable usage of macos speechanalyzer?
Hmm, I don’t think that’s exposed through a web api yet. Will probably have to do it when this becomes an Electron desktop app (soon).
This is now on the roadmap!
I hate Descript, not because of the cost but because of their dark patterns. (I learned recently that their CEO is the former CEO of Groupon, which explains a lot.) So I'd love something like this. But I need to edit an audio-only podcast. Is there a reason that it only accepts video files?
I think you can use ffmpeg to turn your audio into a video file
Yes, you can. Just pass in an image
Now live on ProductHunt! Please support the launch :) https://www.producthunt.com/products/rescript-edit-videos-li...