The best Hacker News stories from Show from the past day
Latest posts:
Show HN: I Turned DOOM into a Transformer
Instead of trying to run DOOM on things, I have instead decided to use DOOM (many, many, many doom games actually) to create a transformer. For a lot of the math required I essentially had to use DOOM game mechanics and voodoo dolls (DOOM is turing complete btw) and exchange them with the actual operations, essentially creating an alternative to NumPy (a 65000x slower one). Thus, I am proud to present DOOMFORMER. Here are a list of things DOOMFORMER can do :<p>It can classify MNIST digits, even if you write these digits on the map!<p>It can talk(kinda)<p>You can use it to make decisions, so, it can play DOOM. Yes! You can play DOOM with thousands of DOOM games now! The marine's mind can be controlled by DOOM, playing on a map generated by DOOMFORMER. Truly, the possibilities are endless, and even larger-scale training of DOOMFORMER could be done on a measly T4.<p>If anyone would like to help me scale this up and make crazier things, I'd love to work together.
Show HN: A working Nintendo in ~1GB of Qwen weights
Show HN: A working Nintendo in ~1GB of Qwen weights
Show HN: A working Nintendo in ~1GB of Qwen weights
Show HN: Squishy 3D helpers powered by Three.js
Show HN: Overtone – Transcribe, correct and cut captions for any screen shape
Hi HN, I'm Ruud, co-founder of Lont. We turn company emails into personalized videos.<p>Video captions cost us manual work twice: fixing misheard words, then fixing layout, like lines ending on “the”.<p>Overtone uses multiple models to listen to the audio: gpt-transcribe, Azure STT, and Elevenlabs Scribe. When they return conflicting results, Overtone sends small clips to the models again, and then lets an LLM decide what was actually said.<p>The LLM is like a human doing a sanity check on the transcription model. Humans would easily recognize a word to be different based on context. The LLM basically does the same, using an optional context prompt + its world knowledge. Because of this the Word Error Rate (WER) on many test files is lower when using Overtone, compared to just using a model without it.<p>Sure, the cost is higher because we use up to 3 different STT models + an LLM, but it still removes mistakes, reducing the time needed to fix captions to a minimum, and reducing the chance you'll miss something. For us it's worth it. For ultra high volume use cases it may not be.<p>The rules-based cutter nicely cuts the transcript into captions that look good, depending on screen size: 16:9 is different from 9:16, and the Overtone Format we created supports any type really. Pauses become breaks, captions are grouped per speaker, and names/numbers are never spread out over multiple captions.<p>How to add these captions to your videos? A videojs plugin can do this automatically, so you don't burn in your captions. The plugin lets you cycle between styled captions, plain captions, and off. You can burn them into an MP4 file with overtone burn.<p>Quickly try it out just the cutter? Try:<p><pre><code> npx overtone-captions recut yoursub.srt
</code></pre>
Transcribing audio with API keys works too. Use --dry-run to show a cost estimate before you send.<p>We open-sourced this first version (MIT) because we believe many people can benefit from this, especially if your captions just need to be good. There's plenty of room to improve though, so we've made it easy to help contribute to add style packs, languages, and providers/models.<p>You can find a demo at: <a href="https://overtone.lont.ai/" rel="nofollow">https://overtone.lont.ai/</a><p>Find the code and benchmarks here: <a href="https://github.com/lonthq/overtone-captions" rel="nofollow">https://github.com/lonthq/overtone-captions</a>
Show HN: Overtone – Transcribe, correct and cut captions for any screen shape
Hi HN, I'm Ruud, co-founder of Lont. We turn company emails into personalized videos.<p>Video captions cost us manual work twice: fixing misheard words, then fixing layout, like lines ending on “the”.<p>Overtone uses multiple models to listen to the audio: gpt-transcribe, Azure STT, and Elevenlabs Scribe. When they return conflicting results, Overtone sends small clips to the models again, and then lets an LLM decide what was actually said.<p>The LLM is like a human doing a sanity check on the transcription model. Humans would easily recognize a word to be different based on context. The LLM basically does the same, using an optional context prompt + its world knowledge. Because of this the Word Error Rate (WER) on many test files is lower when using Overtone, compared to just using a model without it.<p>Sure, the cost is higher because we use up to 3 different STT models + an LLM, but it still removes mistakes, reducing the time needed to fix captions to a minimum, and reducing the chance you'll miss something. For us it's worth it. For ultra high volume use cases it may not be.<p>The rules-based cutter nicely cuts the transcript into captions that look good, depending on screen size: 16:9 is different from 9:16, and the Overtone Format we created supports any type really. Pauses become breaks, captions are grouped per speaker, and names/numbers are never spread out over multiple captions.<p>How to add these captions to your videos? A videojs plugin can do this automatically, so you don't burn in your captions. The plugin lets you cycle between styled captions, plain captions, and off. You can burn them into an MP4 file with overtone burn.<p>Quickly try it out just the cutter? Try:<p><pre><code> npx overtone-captions recut yoursub.srt
</code></pre>
Transcribing audio with API keys works too. Use --dry-run to show a cost estimate before you send.<p>We open-sourced this first version (MIT) because we believe many people can benefit from this, especially if your captions just need to be good. There's plenty of room to improve though, so we've made it easy to help contribute to add style packs, languages, and providers/models.<p>You can find a demo at: <a href="https://overtone.lont.ai/" rel="nofollow">https://overtone.lont.ai/</a><p>Find the code and benchmarks here: <a href="https://github.com/lonthq/overtone-captions" rel="nofollow">https://github.com/lonthq/overtone-captions</a>
Show HN: A proof-of-concept formalization of the IRS federal tax forms
The linked repo is 100% vibe coded, but it was built on a handful of principles:<p>- Most tax software derives trust from the party who maintains it (e.g. Intuit, the IRS, etc.)<p>- Vibe coded software lacks this trustworthiness.<p>- Therefore any formalization should build in verifiability and mechanisms that make it possible for users/maintainers to trace claims/interpretations back to authoritative source documents.<p>It's very much a back-burner passive project right now (do not use this for anything real!), but I feel strongly that something like this ought to exist for the public good. I built it to get a sense of what the shape of what such a thing might be.
Show HN: A proof-of-concept formalization of the IRS federal tax forms
The linked repo is 100% vibe coded, but it was built on a handful of principles:<p>- Most tax software derives trust from the party who maintains it (e.g. Intuit, the IRS, etc.)<p>- Vibe coded software lacks this trustworthiness.<p>- Therefore any formalization should build in verifiability and mechanisms that make it possible for users/maintainers to trace claims/interpretations back to authoritative source documents.<p>It's very much a back-burner passive project right now (do not use this for anything real!), but I feel strongly that something like this ought to exist for the public good. I built it to get a sense of what the shape of what such a thing might be.
Show HN: A proof-of-concept formalization of the IRS federal tax forms
The linked repo is 100% vibe coded, but it was built on a handful of principles:<p>- Most tax software derives trust from the party who maintains it (e.g. Intuit, the IRS, etc.)<p>- Vibe coded software lacks this trustworthiness.<p>- Therefore any formalization should build in verifiability and mechanisms that make it possible for users/maintainers to trace claims/interpretations back to authoritative source documents.<p>It's very much a back-burner passive project right now (do not use this for anything real!), but I feel strongly that something like this ought to exist for the public good. I built it to get a sense of what the shape of what such a thing might be.
Show HN: UK Farming Simulator
Show HN: Auto Dismiss Mac OS "Allow <app> to find devices on local networks?"
This popup has been a constant gripe of mine for years. I almost NEVER want to allow this permission and there is no system setting to disable it globally and prevent the popup from interrupting your typing, etc. This tiny swift app I built will autoclose it in 1/4th of a second and leave behind a little notification that it did so. Would love to hear feedback!
Show HN: Auto Dismiss Mac OS "Allow <app> to find devices on local networks?"
This popup has been a constant gripe of mine for years. I almost NEVER want to allow this permission and there is no system setting to disable it globally and prevent the popup from interrupting your typing, etc. This tiny swift app I built will autoclose it in 1/4th of a second and leave behind a little notification that it did so. Would love to hear feedback!
Show HN: Auto Dismiss Mac OS "Allow <app> to find devices on local networks?"
This popup has been a constant gripe of mine for years. I almost NEVER want to allow this permission and there is no system setting to disable it globally and prevent the popup from interrupting your typing, etc. This tiny swift app I built will autoclose it in 1/4th of a second and leave behind a little notification that it did so. Would love to hear feedback!
Show HN: Popkern – open-source CAD kernel with G2 corner blends, browser IDE
Show HN: A Lego-like composable Framework for RSI
Show HN: Yavashark: I've Built a JavaScript Engine Every Day for 950 Days
Feel free to give me feedback or discuss what you have on your mind. I had lots of fun building this, so I hope you have too reading this!
Show HN: Yavashark: I've Built a JavaScript Engine Every Day for 950 Days
Feel free to give me feedback or discuss what you have on your mind. I had lots of fun building this, so I hope you have too reading this!
Show HN: Nasscad – local-first 3D CAD for browser and Windows
Show HN: Nasscad – local-first 3D CAD for browser and Windows