A recurring beep from an unpacked box room was difficult to trace by ear, so Simon Willison used Claude Code to build Beep Locator: a web app that turns phones, tablets, and computers into networked microphones. The Python server coordinates listener devices through WebSockets, while browser code records recent audio and detects candidate beeps using frequency analysis. When one is detected, a device plays a chirp that helps align the recordings; the server then compares the sound’s arrival across listeners to estimate its location. Devices can also play sounds to help determine their positions, though this proved unreliable enough that Willison added a mode for manually placing them roughly on a map.
The app supports clapping as a test, playback of recordings to identify the target sound, and confirmation of which beep to track. It then uses repeated detections to refine the estimated location and suggest where another listener might improve accuracy. In use, it narrowed the beep’s source to within about 40 centimetres, leading Willison to an old carbon monoxide detector with depleted batteries. The project draws on the 2007 BeepBeep research paper on acoustic ranging with ordinary devices. Its broader demonstration is that a coding agent can gain practical “ears” by analyzing microphone audio through code: Claude could inspect sound data, update listener software live, and iterate on the system while it ran.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.
Researchers tested 16 LLMs on all 55 TIS-100 puzzles, giving each one attempt and time to refine its code. The models found 133 new performance tradeoffs across 18 puzzles and set three outright cycle records, though human players improved on most solutions within days. (tis-100.wincon.dev)
Australia is considering a new criminal offence for AI companies that fail to take reasonable steps to prevent harm caused by their models. The government is also weighing an AI safety watchdog with power to impose significant fines, following a hack of a Medicare statistics portal involving OpenAI. (abc.net.au)
AI is increasingly becoming a gateway through which people find information. Audiences still resist having AI produce their news. (niemanlab.org)
Nvidia holds a central position in the AI industry, making disruption to the company potentially catastrophic for the wider ecosystem. The race for advanced AI is also intensifying economic and geopolitical conflict, while concentrated corporate power and infrastructure face growing political resistance. (france24.com)
LLM inference can require reading about 26 GB of model weights to generate each token from a 13-billion-parameter model. The explanation shows how model size drives memory bandwidth demands during text generation. (youtube.com)
Deletion Tuning matches full fine-tuning across three LLM post-training methods by zeroing fewer than 0.05% of model weights. Its compact masks also preserve prior-task accuracy better than full fine-tuning, LoRA, and random sparse tuning. (arxiv.org)
Today's best Hacker News stories, summarized and screenshotted, one email a day.