Alternatives
Products that do what Magenta Real-Time Music Generation Locally on iPhone, Without the GPU does
Last Thursday, Deepmind released Magenta Realtime 2 , an open source music generation model. They said it could run on Mac, but not iPhone. As a v̵i̵b̵e̵ ̵c̵o̵d̵i̵n̵g̵ ̵a̵d̵d̵i̵c̵t̵ agentic AI maxxi and person who has melted iPhones before (link at bottom), I took that as a personal challenge and made it my weekend project. On Saturday, I got it to run for 10min straight on an iPhone 12 Pro from 2020 without melting the phone or - shockingly - touching the GPU. How? I chopped the model up into 5 pieces and set them each to run on different parts of Apple's system on a chip (SoC). My past…
- 1
- 2
I trained a 125M-parameter transformer to autocomplete piano performances in real time (~108 notes/sec on an iPhone 15). The idea is basically GitHub Copilot or Tabnine, except instead of prompting it with code, you prompt it by playing a few notes on a MIDI piano. The model then continues what you played, entirely on-device. The app is free if anyone wants to try it. Happy to answer questions about the model, training, Core ML, or the many things that didn't work.
17d ago · simedw.com
- 3IB
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…
Oct 2025 · ambisounds.app
- 4RT
I ported Microsoft's TRELLIS.2 (4B parameter image-to-3D model) to run on Apple Silicon via PyTorch MPS. The original requires CUDA with flash_attn, nvdiffrast, and custom sparse convolution kernels: none of which work on Mac. I replaced the CUDA-specific ops with pure-PyTorch alternatives: a gather-scatter sparse 3D convolution, SDPA attention for sparse transformers, and a Python-based mesh extraction replacing CUDA hashmap operations. Total changes are a few hundred lines across 9 files. Generates ~400K vertex meshes from single photos in about 3.5 minutes on M4 Pro (24GB). Not as fast as…
Apr 2026 · github.com
- 5AN
I built this free, no-nonsense white noise app. I know there are plenty of them out there already, but I wanted to make something beautiful and easy-to-use, without logins or ads or in-app purchases or any of the other stuff a lot of them have. I appeared on The Economist's The Intelligence podcast [0] this week talking about Ambiphone and another ambient music project, Ambient ScotRail Beats [1] - I'm on at about 17:30 There's a big selection of music and sounds already but I'm always adding more - if there's anything you'd like to see added, let me know! [0]…
2024 · ambiph.one
- 6RS
2024 · apps.apple.com
- 7
- 8
- 9AS
Hi HN, author here. SHARP is Apple's recent single-image 3D Gaussian splatting model (https://arxiv.org/abs/2512.10685). Their reference code is PyTorch + a pretty heavy pipeline; I wanted to see if it could run in a browser with no server hop, so I exported the predictor to ONNX and ran it via onnxruntime-web with the WebGPU EP. What works: drop in an image, get a .ply you can download or preview live, all on your machine — your image never leaves the tab. The model is large (~2.4 GB sidecar) so first load is slow on a cold cache, but inference itself is a few seconds on…
May 2026 · github.com
- 10
- 11
- 12
- 13
- 14
- 15
- 16IM
I made this iPhone app called "Affine Tuning" which is a collection of body-interactive musical pieces: https://affinetuning.com/ It uses body tracking (Motion Capture from RealityKit) to shape dynamic compositions. The app is completely free and will be updated in the future with more songs and other interactive features.
2022
- 17
- 18AG
Hey folks! I've spent the last 8 months building a VST plugin called "The Infinite Crate" on top of Magenta's Lyria RealTime music model on the Gemini API. Doug Eck's research group has been building MusicLM and variants since 2017 (https://www.youtube.com/watch?v=yz-fHidp1M8) I bridged a JUCE/C++ audio processing foundation with a React/Typescript UI, using Zustand to sync state. The plugin streams down generated audio from the Lyria RealTime model in the Gemini API via websockets You can try it as a VST3 on Mac/Windows or AU/Standalone on Mac. It works in…
2025
- 19IB
Hey HN, I built Presync.io to simplify plugin management for music producers on macOS. Presync lets you easily discover, install, and manage music production plugins. Read it as "Steam for music production software". What it has now: - One-Click Installation - 110+ Free Plugins - Video Reviews and Tutorials for plugins (in progress) - Management Stuff: delete for any plugin in your system, MVP of updates (very early stage) Backstory: I love music production and development. This past spring, I started Presync to move away from the endless methods of reading/writing PDFs and Excel files.…
2024 · presync.io
- 20
Mirror and control your real iPhone
Jul 2026 · mirrorue.xyz
- 21TF
I’d originally launched my app: Private LLM[1][2] on HN around 10 months ago, with a single RedPajama Chat 3B model. The app has come a long way since then. About a month ago, I added support for 4-bit OmniQuant quantized Mixtral 8x7B Instruct model, and it seems to outperform Q4 models at inference speed and Q8 models at text generation quality, while consuming only about 24GB of RAM[3] at 8k context length. The trick is: a) to use a better quantization algorithm and b) to use unquantized embeddings and the MoE gates (the overhead is quite small). Other notable features include many more…
2024
- 22VB
I kept seeing people making macOS apps for their one-off use cases and automations. I thought surely the hassle of paying for an Apple dev account and fighting with code signing and notarization would be a deterrent for most people. Viberunner is a container environment that lets you build macOS apps with simple prompts. You can run the apps in a new tab, a new window, or even as a system menu bar app. You can create things like: - a clipboard manager - an image redactor - a note-taking tool that saves a daily note to a folder on your desktop - a desktop organizer that cleans up your…
2025 · viberunner.me
- 23LT
Hey HN, I've used Generative.fm for years and love it, but I always wanted to just describe what I was in the mood for instead of scrolling through presets. So I built this. You type a text description of anything - from "mountain sunrise" to "neon city" - and it generates a procedural/ambient stream matching that mood. It runs locally, no account, no tracking, no ads. Under the hood it's a custom synthesizer driven by sentence embeddings, not a generative AI model (although you can choose to use one!) — so there's no GPU, no API calls, and it starts playing almost instantly. The whole…
Feb 2026 · latentscore.com
- 24MH
We're a small group competing at a local StartupWeekend (http://startupweekend.org) and were hoping to get feedback on our app idea. We're very close on a working prototype for android, using accelerometer data to determine music that would match a user's intensity level. Most of us are hackers, working away at code and design/copy is admittedly a weak point. We're also working on our market validation at the moment. Our website is at http://movegroove.github.com. Any and all feedback is appreciated! Thanks for your time.
2011
Ranked by how close each launch is in meaning, then by votes. Refine with a description →