2025년 10월 12일의 Show HN

I’ve always liked having background noise while working or falling asleep, but I got frustrated that most “white noise” or ambient sound apps are either paywalled, stuffed with ads, or try to upsell subscriptions for basic features.

So I made Ambi, a small iOS app with a clean interface and a set of freely available ambient sounds — rain, waves, wind, birds, that sort of thing. You can mix them, adjust volume levels, and just let it play all night or while you work. Everything works offline and there are no hidden catches.

It’s something I built for myself first, but I figured others might find it useful too. Feedback, bugs, and suggestions are all welcome.

https://apps.apple.com/app/ambi-white-noise-sleep-sounds/id6...

i made an esoteric programming language which i call spellscript. every program is a "spell" written in a "grimoire," and you have to use keywords like summon, enchant, inscribe, and conjure.

it's literally read like a spellbook because the syntax consists of all natural language, and newlines are optional. your code can now be an essay, like everybody wants!

for example, if you want to print something, you'd write: `begin the grimoire. inscribe whispers of "hello, world!". close the grimoire.`

it has variables, dynamic typing, arrays, functions, conditionals, loops, string manipulation, array manipulation, type conversion, and user input, among other (listed in the docs!)

but why? i wanted to see how far you could push natural language syntax while still being parseable. most esolangs are intentionally obtuse (BF, Malbolge), but i wanted something that's weird but readable, like you're reading instructions from a spellbook, which makes it incredibly easy to read and understand. like an anti-esolang? hmm...

github: https://github.com/sirbread/spellscript

docs: https://github.com/sirbread/spellscript/blob/main/resources/...

Alt link: https://mrchristmas.com/products/santas-magical-telephone

Video demo: https://www.youtube.com/watch?v=0z7QJxZWFQg

The first time I talked with AI santa and it responded with a joke I was HOOKED. The fun/nonsense doesn't click until you try it yourself. What's even more exciting is you can build it yourself:

libpeer: https://github.com/sepfy/libpeer

pion: https://github.com/pion/webrtc

Then go do all your fun logic in your Pion server. Connect to any Voice AI provider, or roll your own via Open Source. Anything is possible.

If you have questions or hit any roadblocks I would love to help you. I have lots of hardware snippets on my GitHub: https://github.com/sean-der.

I made Sober not Sorry, a clean and simple iOS tracker to help quit bad habits — alcohol, smoking, or anything else. Open it → log your start date → it tracks your streak. No ads, no sign-up, no subscriptions. I built it because most “sober” apps felt cluttered or too motivational. This one just quietly tracks progress and stays out of the way. Would love feedback — especially from people who use streak or sobriety trackers.

I listen to online radio all the time while working, but none of the web based apps I found fit my needs.

So I built RadioPuppy.com - https://radiopuppy.com

It’s a lightweight web app that lets you quickly search thousands of online streaming radio stations and save your favourites.

I’m using the excellent https://www.radio-browser.info API as the data source for stations.

The tech stack is: Laravel 12, Inertia.js, React, Redis, and PostgreSQL.

Future features I am thinking of...

- Let users upload their own stream URLs, so if something isn’t available via the API, they can still save and listen to it. - Plot stations on a map to browse by location. - Play History: Keep track of stations you’ve listened to,

Would love any constructive feedback, thanks!

I built Hyprvoice, a small tool that lets you type with your voice on Wayland — no X11, no kludgy bridges. Press a key, talk, and your words appear wherever your cursor is.

It’s written in Go, uses PipeWire for audio, and talks directly to the compositor. The workflow is simple:

press key → start recording

press again → stop and inject text

Notifications show recording/transcribing states, and it works with multiple backends — currently OpenAI Whisper, with whisper.cpp (local/offline) in progress.

Everything runs through a lightweight daemon with IPC over a unix socket. Text injection uses wl-clipboard + wtype with clipboard restore fallback.

Install (Arch/AUR):

yay -S hyprvoice-bin systemctl --user enable --now hyprvoice.service

Then add something like this to Hyprland:

bind = SUPER, R, exec, hyprvoice toggle

Repo: https://github.com/leonardotrapani/hyprvoice

It’s beta but fully usable. I’d love feedback, especially from people running Wayland full-time or maintaining compositors.

I integrated a remote GPU execution backend into PyTorch through the same system that custom hardware accelerators get integrated into PyTorch. You can create a remote machine and obtain its CUDA device whenever you want to create or move tensors onto the remote GPU.

  machine = mycelya_torch.RemoteMachine("modal", "A100")
  cuda_device = machine.device("cuda")
  x = torch.randn(1000, 1000, device=cuda_device)
  y = torch.randn(1000, 1000).to(cuda_device)

I made it reasonably performant by having most operations dispatch asynchronously whenever possible. For cases where slow performance is unavoidable such as uploading many GB of weights onto the GPU, I added a decorator that can be applied to functions to turn it into a remotely executed function. For the most part, the function should behave the same with or without the decorator; the main difference is whether the function code is executed locally or remotely.

  import mycelya_torch
  from transformers import AutoModelForCausalLM, AutoTokenizer

  @mycelya_torch.remote
  def load_model(model_name: str):
      tokenizer = AutoTokenizer.from_pretrained(model_name)
      model = AutoModelForCausalLM.from_pretrained(
          model_name, torch_dtype="auto", device_map="auto"
      )
      return model, tokenizer

You can only use it with Modal as the cloud provider right now, and it's free to use with their free monthly credits. I appreciate any feedback and bug reports :)

Got tired of all the ads, cookies etc when reading my news on my phone or web browser. Created a service that access my favorite rss news feeds and delivers them to my kindle directly, no ads or distractions. The news articles are also summarized for when I don't have enough time to consume my news.

I built Videoyards, a browser-based screen recording and editing tool made for indie hackers and SaaS founders who want to create clean, professional demo videos without wasting time switching tools.

It works directly via a Chrome extension, no downloads, no setup. You can record your screen, camera, and mic, then instantly edit with auto zooms, cursor highlights, and silence removal right in your browser & Exports are available in HD or 4K 60FPS.

The main benefit is speed. you can record, polish, and share a demo in minutes instead of hours. It’s built to help you show your product clearly, make your demos look professional, and save time so you can focus on building instead of editing.

Free plan includes 1-minute HD recordings with watermark. Early supporters can grab a $39.99 lifetime deal for unlimited 4K recordings, camera/mic support, and lifetime updates. Currently i had 200+ users using it and got an rating of 5.0. Still in the beta version and trying to make it better.

Would love your feedback on the recording speed, editing flow, and overall UX compared to Loom or Screen Studio.

I put together some interactive notebooks to help people get started with Linux Bash scripting.

Runs in the browser, no setup needed.

Hey HN,

I’m building Databite, an open-source set of tools to help developers build integrations faster and with ease.

If you’ve ever built a SaaS product or an AI agent that needed to update a CRM record, sync data between apps, or trigger a workflow in a third-party tool, you’ve probably spent way too much time writing glue code. Databite is here to be that glue.

With Databite you can: - Access prebuilt connectors using the databite/connectors package. - Create custom connectors to services of your choice like Slack, Notion, HubSpot, Google Sheets, etc. using our databite/build package. - Embed one click authentication flows to your React projects using our databite/connect package . - Schedule syncs and mange connections using our databite/engine package. - Provide docs and have AI build new connectors for you using our databite/ai package.

The idea is to make building integrations as easy as npm installing a library.

This is especially useful if you’re building AI agents that need large action libraries. Databite can serve as the “integration layer” for your agent to perform real-world tasks.

We’re still very early (wouldn't recommend using the libraries for anything serious just yet, sharing early or I might just never do it (: ). The goal is to start building a community of interested contributors. I believe given the nature and scale of the problem, integrations can only be truly solved by a vibrant open source project (we're under the MIT license which allows almost every imaginable use case).

You can find out more on our website (https://databite.dev). Or watch us on Github and get notified on our progress (https://github.com/DatabiteDev/databite)

As the weather here in the Pacific Northwest is getting a big damp, here's an app called BLE Sync Cycle (currently Linux-based) that I wrote that allows me to interactively cycle indoors in front of a big-screen TV/monitor (similar to Zwift or Rouvy).

Additional details on the GitHub site: check out the project wiki (https://github.com/richbl/go-ble-sync-cycle/wiki) and roadmap (https://github.com/users/richbl/projects/4).

Python has lacked a batteries-included HTTP library that would have both async and sync clients. Httpx (httpcore), which has offered this, is unfortunately pretty much unmaintained and suffering from huge perf issues (https://github.com/encode/httpx/issues/3215).

I built pyreqwest HTTP client for Python that is fully Rust based on top of reqwest. It includes all features reqwest offers plus some more. Also including unit testing utilities (mocking, ASGI app support). Go check https://github.com/MarkusSintonen/pyreqwest :)

I often look for domains when brainstorming new projects, but most domain marketplaces are full of gibberish names or already-taken domains. So I built SeekLex — it preprocesses over 1M aftermarket domains to filter out nonsense, detect meaningful words, and generate creative domain hacks when your preferred names are taken.

The goal is to make domain discovery smarter and faster, using automation and (soon) AI. I’m planning to add an API, MCP integration, and an LLM agent to help you find the best domain for your next project.

Would love your feedback and ideas.

The game of binary rain. Totally inspired by The Matrix and cmatrix specifically. It's just a simplified typing game. Just two keys for control, 0 and 1. Type the digits in sequence within the falling strains to clear them before they reach bottom.

Made with Nim. Zero dependencies, under 200KB compiled, crazy fast compile time (under 2 seconds on my PC).

I've been exploring how diffusion models, which have transformed image and video generation, can be applied to text generation. The math behind discrete diffusion models can be a bit intimidating at first, so I put together an annotated Jupyter Notebook that walks through the theory and builds a character-level discrete diffusion GPT, adapted from Andrej Karpathy's baby GPT. Instead of generating text autoregressively (left to right), the model learns to denoise corrupted text sequences in parallel.

The notebook covers the mathematical framework, noising process for discrete tokens, and a working implementation trained on Shakespeare's text.

GitHub: https://github.com/ash80/diffusion-gpt

Open in Google Colab: https://colab.research.google.com/github/ash80/diffusion-gpt...

Feedback and suggestions are very welcome.

I built an app with a Duolingo-style structured learning system where users can generate full courses from photos, documents, or typed input. Each course includes quizzes, flashcards, and notes, and users can also turn their notes into an AI-generated podcast for audio learning.

It’s designed to help students and anyone who wants to learn a topic study smarter and retain more through interactive lessons, quizzes, and podcasts tailored to their learning style.

Built with React Native Expo.

https://youtu.be/jsRk1vN1xNA?si=iMGZFhj_N2iwBp5n

I built a simple tool to convert PNG/JPG/WebP images to PDF without any uploads. Everything processes locally in your browser - drag to reorder pages, adjust layouts, and export instantly. No accounts, no watermarks, no file size limits beyond your device's memory.

https://pngtopdf.cc

The main benefit is privacy: your files never leave your computer. I was tired of uploading sensitive documents to random conversion sites, so I made this client-side alternative using JavaScript.

It's free and works offline once loaded. Feedback welcome!

I built a bike trainer-powered arcade browser game for parties, bike fairs and other bike-related events. The game is very easy: Every X seconds the target watts increase by Y. The person who can keep up the longest, wins. Hope you like it :)