2025年7月13日 の Show HN

I'm in my 4th year living in Denmark as an expat, and I finally decided it’s time to properly learn Danish. I do have a Danish girlfriend, after all. One way I’ve been practicing is by trying to text only in Danish, but I often find myself stuck. I start my message in Danish, then hit a wall because I don’t know a word or how to fit something naturally into the sentence.

Especially in those cases, I used to give up and translate the entire message from English, which kind of defeats the purpose and interrupts the learning process.

So I started prompting GPT. I’d write my message with wildcards or notes for the parts I didn’t know, and it would return a corrected version. That worked well, but reusing the prompt each time became tedious.

So I built a wrapper around it.

Now I can type in the target language, mark unclear parts with curly braces {like this}, and get an instant corrected version with explanations. I also added a history feature so I can review what I got wrong, and I plan to build more on that soon (eg. summary of areas or words to review).

This app is for language learners who want to practice writing without feeling insecure about mistakes or breaking their flow by switching to a translator.

I hope you find it useful!

Hey HN!

I'm a huge fan of Raycast, but as a Linux user, I was always disappointed it wasn't available on my main OS. This summer, I decided to just build it myself. This project has the goal of being interoperable with Raycast itself, including a majority of the extensions.

It's built with Tauri and Rust on the backend, with a Svelte frontend. The biggest challenge was getting it to run existing Raycast extensions, which required building a custom React renderer as well as making a custom API.

I also wrote a quick post, which I hope to expand on in the future, about this project. You can find it here: https://byteatatime.dev/posts/recreating-raycast

The project is still in alpha, but I'm excited to share it and get your feedback.

This is a spiritual adaptation of oil.nvim for vscode. The main idea is you edit the filesystem by editing the current directory listing's text buffer. For example, if I want to rename a file, I just rename it in the listing file. This is extremely powerful because it translates all of your text-editing skills immediately into file editing capabilities.

Some features:

* Create/rename/move/delete files by editing the current directory listing's textbuffer

* Filter using glob pattern

* Trash and undo support

* Works even in remote-ssh workspaces

* Works across multiple vscode windows

Hey HN! I got nerd-sniped by Bloom Filters this weekend, specifically for searching datasets with high "cardinality" (number of unique items).

They're an _amazing_ data structure that, at a fixed size, tracks potential set membership. That means unlike normal b-tree indexes, they don't grow with the number of unique items in the dataset.

This makes them great for "needle in a haystack" search (logs, document) as implementations like VictoriaMetrics and Bing's BitFunnel show. I've used them in the past, but they've never been center-stage in my projects.

I wanted high cardinality keyword search for ANOTHER project... and, well, down the yak-shaving rabbit hole we go!

BloomSearch brings this into an extensible Go package:

- Very memory efficient via bloom filters and streaming row scans

- DataStore and MetaStore interfaces for any backend (can be same or separate)

- Hierarchical pruning via partitions, minmax indexes, and of course bloom filters

- Search by field, token, or field:token with complex combinators

- Disaggregated storage and compute for unbound ingest and query throughput

And of course, you know I had to make a custom file format ^-^ (FILE_FORMAT.MD)

BloomSearch is optimized for massive concurrency, arbitrary cardinality and dataset size, and super low memory usage. There's still a lot on the table too in terms of size and performance optimizations, but I'm already super pleased with it. With distributed query processing I'm targeting >100B rows/s over large datasets.

I'm also excited to replace our big logging bill ~$0.03/GB for log storage with infinite retention and guilt-free querying :P

I'm lost between ChatGPT vs Claude vs Gemini... which subscriptions to take? With Cursor and all these specific AI tools, I just wanted one simple chat app where I can use any model and pay only when I use it.

Couldn't find one, so I built one.

Pay only for what you use. Your prompts and docs, knowledge bases work with every model - no more copy-pasting between apps.

Started as a personal project, but thought someone else might benefit from this too.

https://prismharmony.com/chat

What do you think?

I built a symbolic dream interpretation engine that runs fully in the browser no AI, no backend, no cloud. Just vanilla JavaScript and a hand-mapped keyword system. It scans dream text, matches symbolic keywords to over 300 custom meanings, using deterministic logic. All data stays in localStorage no tracking, no server. Try it here: https://dino-nuggies45.github.io/Dream-Interpreter/

Built to explore subconscious patterns and emotional systems without machine learning. Feedback welcome!

HNping lets you set a reminder for a HN post. You get a web push browser notification when it's time, and clicking it takes you back to the post. That's it.

I built HNping because I kept stumbling on HN posts where the discussion still had to get going. Wanted to revisit, but didn't want to create even more bookmarks/etc I'll just forget about. So I created a 'remind me later' tool (like the reddit bot) to fix this for myself.

To use it: go to hnping.com, enable notifications, and drag the bookmarklet to your bookmarks bar. Then click it on any HN post to set a reminder (5 minutes to 1 week). No personal info needed - you just get a UUID that serves as your account.

I tried to make it as simple as possible.

It's built on a Cloudflare Worker with D1 for data storage and uses Firebase Cloud Messaging for push notifications.

I'm still iterating on it and would love any feedback or suggestions. I'm considering creating a Chrome extension when the need is there. Thanks for checking it out!

Hey HN, I love creating, and I've been fascinated by AI audio tools. But as a creator, trying to actually use them felt like a trap.

First, wrestling with raw TTS APIs to create something that doesn't sound like a robot reading a phone book is an art form in itself. It takes forever to get the pacing and tone right.

Second, and this was the real deal-breaker: even if you create a masterpiece, platforms like YouTube feel like they're just waiting to demonetize you or slap a restriction on your work because it's "AI-generated." It feels like you're building on someone else's land, and they can change the rules overnight. It's soul-crushing.

I wanted a sanctuary for audio creators. A place that’s more like a recording studio in your pocket than a frustrating API wrapper. A platform where you can focus on your story, not on fighting the tools or the distribution algorithm.

So, I built Sohri (https://sohri.ai). Sohri is designed to turn your short stories into polished, binge-able audio episodes, handling the boring parts so you can be the director. During testing, I was surprised to find that for many shorter pieces, the AI narration actually felt more engaging than my own quick recordings because it never gets tired or stumbles.

It’s free to use, and the core features will always be free. My goal is to build a place where creators can experiment, share their work, and eventually earn a living without fear. I’m here all day to answer any questions. I'd be incredibly grateful for your feedback—try listening to a story or creating your own and let me know what you think.

I made a small CLI tool called `c0admin`. Runs locally using your own Gemini API key. No signup, no server, no tracking.

You just run `c0admin` in your terminal, and it gives you suggestions interactively.

Antilisp is a Lisp designed for code generation in non-lisp languages. The interpreter is written in RPython, and the language is designed for easy adoption by non-lispers. The project is still young, but the language can be played with if you don't mind missing some important features like modules and pattern matching.

I have not published the source code yet, because I am not completely sure how to base a sustainable business on this project and I don't want to risk having to rugpull after raising expectations.

It provides zero-dependency, memory-safe operations for working with torrent files, bencode data, and magnet links.

We needed a speaker diarization solution that could run every few seconds alongside transcription on iOS and macOS. But native Swift support was either limited or locked behind paid licenses. Since diarization is a common need in speech-to-text workflows, we decided to open source our work and give back to the community.

We initially tried sherpa-onnx, which works, but running both diarization and transcription models slowed down older devices. CPU-only inference just isn’t ideal for near real-time workloads, so we wanted the option to offload segmentation and speaker embedding to the GPU or ANE. Supporting M1 Macs in particular meant pushing more of the workload to the ANE.

Instead of shoehorning the ONNX model into CoreML with C++, we converted the original PyTorch models directly to CoreML. This approach required some monkey-patching in the PyTorch and pyannote code, but the initial benchmarks look promising.

We’d love feedback! We're currently working on adding VAD and integrating Parakeet for transcription, but still wrestling with CoreML model conversion.

Hey HN,

We built AutoAI – an AI tool that tells you what's wrong with your car in plain English.

Just enter:

Your car’s make/model/year

The OBD2 error codes (optional) (like P0420, P0171, etc.)

Any symptoms you're noticing (e.g. “rough idle” or “weird sound when starting”)

And we’ll tell you:

The most likely issue

How to verify it yourself

Whether it’s a DIY fix or shop-worthy

No more endless Googling or forum-hopping. Built for car owners, tinkerers, and pros who want fast, reliable answers. Powered by a repair-trained AI using real-world automotive data.

We’re trying to make diagnostics smarter, not replace your mechanic – just make you way more informed before spending money.

Would love feedback or crazy edge-case inputs to improve it.

I call my Portfolio Website "The Best Terminal-Inspired Portfolio Website on The Internet™"

As one of the games, I decided to play around and add this - just use "gameoflife" as the command and try it out, I'd love to hear thoughts :)

I built this because I was tired of scrolling through hundreds of lines of AWS CLI JSON output just to find instance status. hawk brings pandas-like operations (select, group_by, aggregations) to CLI with unified syntax across JSON/YAML/CSV.

Key features: - Instant data structure overview with `| info` - Same query syntax for all formats - Built in Rust for speed and single-binary distribution

Would love feedback from the community!

We wanted to know how well LLMs can predict what you think and put them to the test in a game of codenmaes! Grab a friend and play in two teams, each consisting of one human and one LLM.

Do you think LLMs can grok your clues?

I wanted a way to limit what channels on youtube my kids could watch, so I built this over the weekend with Claude. Whats nice is it still works even though we're blocking youtube and youtube-kids through PiHole.

No idea what I'm going to do with it from here, but thought it was worth a share. Thanks!

Quikfolioz is a curated collection of inspiration, ideas, and ready-to-use portfolio templates designed for developers who want to quickly establish their online presence.

As someone who spends a lot of time on world-building and TTRPGs, I've always found a gap in the tools available. Existing name generators are great at combining syllables based on rules, but they can't capture a feeling or a vibe. You can't ask them for a name that sounds like "an ancient, sunken kingdom." Nomenus is my attempt to solve this. It's an AI-powered tool built around a single prompt, allowing you to describe the name you want in plain, natural language. For example, you can ask it for names for "a grumpy but loyal dwarven blacksmith with a secret love for poetry," and it will generate results based on that specific context, often with a bit of lore attached. The site you see is the first MVP. It's built with Next.js (App Router), Supabase, and runs on Cloudflare Workers. I'm launching it here on HN to get honest feedback from a community that appreciates how products are built. I'm particularly curious about: Is the core concept genuinely more useful than traditional generators? Any thoughts on the UI/UX? What features or generator types should I prioritize next? It's completely free to use. Looking forward to hearing what you think!