Comments URL: https://news.ycombinator.com/item?id=50229413

Points: 1

# Comments: 0

I know there are plenty of time trackers. I wanted something that combined automatic tracking, detailed charts, and cross-device sync in a lightweight app with an interface I enjoyed using. I couldn’t find a free option that fit, so I started building my own.

Hindsight records active apps and window titles while it’s running, without requiring you to start and stop timers. You can explore your activity through daily, weekly, and monthly charts, drill down into individual apps and window titles, and use a yearly heatmap to spot patterns. Categories are customizable.

There’s also an optional screenshot and OCR feature for finding something you previously saw on screen. Both are off by default. When enabled, you can search recognized text and jump back to the corresponding screenshot.

The AI features can generate daily summaries or answer questions about your activity, with links back to the underlying records so you can check the answers. You can use a local model or connect a cloud API; third-party API fees may apply.

Optional sync through Google Drive or WebDAV lets you view activity from Windows and Mac devices separately or together. I’ve also recently added plugin support so other developers can extend the app.

I’d appreciate feedback, particularly on how you’d want to explore your activity history. What would make this useful to you beyond simply showing how many hours you spent in each app?

Happy to answer questions about how it works.

Comments URL: https://news.ycombinator.com/item?id=50229028

Points: 2

# Comments: 0

Alongside the ability to plan routes from a starting song to a final destination, the routes through the music map are navigable too.

There's a few other features like plugins to connect radio stations, or other music sources too which were a big project in terms of getting the plugin structure right to get working with almost any online source now.

Looking for feedback on the music map, how it can be further improved, how intuitive, did it lead to new musical discoveries of your music?

Comments URL: https://news.ycombinator.com/item?id=50228579

Points: 1

# Comments: 0

Comments URL: https://news.ycombinator.com/item?id=50228426

Points: 1

# Comments: 0

I started a small janky passion project called https://privateequityvet.org which used Leaflet.js to map corporately owned veterinary practices in the US. The site gradually took off, was covered on PBS Newshour and now gets 20k uniques a day.

A few months ago I launched IndependentVet.com which aims to be the opposite of the original site, helping pet parents in the US/Canada/UK identify locally owned veterinary practices. Just yesterday I launched native mobile apps that I developed over the course of three intense weeks of evenings and weekends using Opus 5.5. While it's early days, I'm really happy with how it turned out: https://independentvet.com/get

Would love to hear what you think!

Comments URL: https://news.ycombinator.com/item?id=50228314

Points: 3

# Comments: 0

A Slack message, some objectives, a terminal, and at times a browser window or a code file welcome you in each level.

Each level is a mix of 'whose fault it is '- sometimes it's you, sometimes it's your made-up teammate's, and sometimes it's just a bad Monday situation.

It's a real git engine so the answers are judged on the final state of the repo. No LLMs checking it, and no direct command-based solutions.

It is more intended for early devs and students.

Of course AI can solve all your git problems but at the end of the day, fundamentals tell the difference. So git is needed. Even if AI handles git for you, it doesn't hurt to know the whys and hows. And this is where a game like this helps.

Please go and break something. I'd love to know of levels where the commands don't behave as expected. Link: https://git-ing-over-it.vercel.app/

Comments URL: https://news.ycombinator.com/item?id=50228301

Points: 1

# Comments: 1

Comments URL: https://news.ycombinator.com/item?id=50228239

Points: 3

# Comments: 0

Comments URL: https://news.ycombinator.com/item?id=50228115

Points: 4

# Comments: 1

Comments URL: https://news.ycombinator.com/item?id=50228031

Points: 1

# Comments: 0

Comments URL: https://news.ycombinator.com/item?id=50227987

Points: 2

# Comments: 0

Comments URL: https://news.ycombinator.com/item?id=50227984

Points: 2

# Comments: 0

And... Work mode: a shared environment where you and AI agents can browse, research, build, and get real work done together. Live browser sessions, parallel sub-agents, tools, files, components, and interactive results, all in one canvas.

AI is optional (disable in one click). No telemetry. BYOK support and local models.

Website: https://zephium.app Launch video: https://youtu.be/QbSarcWspvw X: https://x.com/CryntaDev

Comments URL: https://news.ycombinator.com/item?id=50227943

Points: 1

# Comments: 0

Comments URL: https://news.ycombinator.com/item?id=50227918

Points: 1

# Comments: 0

Comments URL: https://news.ycombinator.com/item?id=50227632

Points: 1

# Comments: 0

Comments URL: https://news.ycombinator.com/item?id=50227610

Points: 1

# Comments: 0

Comments URL: https://news.ycombinator.com/item?id=50227443

Points: 1

# Comments: 0

Since 2021, I’ve been collecting fully consented voice data from over 20,000 people, 1M audio recordings, and over 1,800 hrs of audio captured to date. For this model, I used a small part of the data, 136 hrs, to finetune Maya1 and expand it’s capabilities.

Hurston-1 is a narrowband TTS model that focuses the output on telephone use because that is still where most of these voices will be heard. I’ve used it to support customers such as Amazon and Salesforce and generated over half-mill revenue with it. Now I’m open sourcing it to help unlock opportunities for others as well.

Limitations: this model doesn’t maintain the exact same voice across prompts and doesn’t support voice cloning. For many use cases, this is fine but maybe not so - if you’re creating podcasts.

This is the first release - I’m already working on the next version with improvements.

I would appreciate, if anyone is interested in this space, download the model from huggingface, check it out and let me know any feedback. Also, open to connecting with others in this space as well.

Comments URL: https://news.ycombinator.com/item?id=50227330

Points: 1

# Comments: 1

Comments URL: https://news.ycombinator.com/item?id=50227243

Points: 1

# Comments: 0

Comments URL: https://news.ycombinator.com/item?id=50227041

Points: 1

# Comments: 0

-randomly: no periodic pattern or undesired ordered structure, to prevent aliasing effects.

-uniformly: no empty regions, clusters... that would unavoidably appear by just using a raw random generator because of natural fluctuations.

In spectral language, one can take the Fourier transform of the point cloud. For a blue noise, it's amplitude should be:

- 0 or small at low frequencies. This ensures long-range order.

- 1 or not too big at high frequencies (short-range disorder).

Hence blue noise: Imagine sending light on the point cloud; in some abstract sense, only the blue (high frequency) will pass through.

Many popular algorithms can sample blue noise: Poisson disks, void and clusters, Lloyd relaxation... but the bluest blue noise (those that show up as a hard 0-1 spectral step) comes from the most basic, brute-force algorithm you can imagine. Take points from a purely random generator, compare the Fourier transform with the 0-1 target, launch gradient descent... and wait. Wait because it has O(N^2) complexity. Well, it turns out in 2D / 3D, advanced fast Fourier techniques can bring it down to N log N, see https://github.com/martiniani-lab/FReSCo (not me), but this is limited to 2D / 3D point clouds.

So, all this just to say that I launched a high-quality but slow blue noise optimiser on the GPU, collected the samples in a reusable database and hope it could interest people from graphics, statistical physics, maybe also quasi-Monte Carlo integration (but this one is just a maybe, simple Sobol sequences are better than blue noise on many tasks) or just curious hackers.

Sampled clouds, all open access: 2D to 32D 1k to 130k points (and up to 30 millions in 2D / 3D thanks to the mentioned fast Fourier speed-up).

Comments URL: https://news.ycombinator.com/item?id=50226642

Points: 1

# Comments: 1