[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"$fombzfcxymd48":3},{"code":4,"message":5,"data":6,"timestamp":41},10000,"success",{"full_name":7,"position":8,"description":9,"language":10,"stars":11,"forks":12,"stars_gained":13,"age_days":14,"hn_threads":15,"one_liner":22,"summary_md":23,"topic":24,"github_topics":27,"license":32,"homepage":33,"updated_at":34,"history":35},"cjpais\u002FHandy",10,"A free, open source, and extensible speech-to-text application that works completely offline.","Rust",34571,3284,88,619,[16],{"id":17,"title":18,"points":19,"comments":20,"posted_at":21},46628397,"Handy – Free open source speech-to-text app",247,110,"2026-01-15","A desktop dictation tool for people who want speech-to-text in any app without sending audio to a cloud service.","Handy is worth a look if you want push-to-talk dictation that lands in whatever text field already has focus, and you are willing to manage a local model download. It is a small utility with a narrow purpose rather than a transcription workbench.\n\n## The problem\n\nYou want to dictate into an editor, chat window or email client without switching apps or uploading recordings. Cloud dictation tools add a step where audio leaves your machine, and lightweight desktop alternatives can be awkward to configure. With Handy, pressing a shortcut records, releasing it runs local speech recognition through Whisper or Parakeet V3, and the text is pasted into the field you were already using. The workflow change is that transcription becomes a keystroke inside the current app instead of a separate service or web page.\n\n## What's different\n\nThe processing really is local: silence is filtered with Silero voice-activity detection, then a chosen model transcribes the clip, with GPU acceleration for Whisper models where available. Parakeet V3 is aimed at CPU-only machines and adds automatic language detection, which matters if your laptop lacks a usable GPU. You also choose between hold-to-record and tap-to-toggle shortcuts, so the tool can fit different input habits rather than forcing one gesture.\n\nThe extension path is unusual for this kind of utility. The README describes Handy as extensible, and the project has a Discord. A commenter in the [launch discussion](https:\u002F\u002Fnews.ycombinator.com\u002Fitem?id=46628397) noted that programming-oriented dictation, such as saying folder paths or git commands, would benefit from dedicated fine-tuning. That is a sensible expectation to calibrate before adopting. The README's description is still one job: transcribe and paste. Anyone expecting the app to understand code context and generate code should treat that as a future experiment, not current behaviour.\n\n## Who should try it\n\nGood fit: someone with repetitive strain or limited typing ability who wants to dictate into existing applications on Windows, macOS or Linux, and is comfortable selecting a model once. The offline design also suits people who cannot send voice data off-device. Poor fit: teams that need shared transcripts, speaker labels or a searchable archive, since the README presents a single-purpose input tool, not a meeting transcription service. It is also a poor first choice if you need dictation of code and command syntax to be exact; the model and the lack of a programming-specific tuning path are the limiting factors.\n\n## Requirements and limitations\n\nYou must grant microphone and accessibility permissions, and on macOS that accessibility grant is what lets the app paste into other fields. Model download is a setup step; behind a proxy or restrictive network you have to fetch the files manually, place them with their exact filenames in the app's models directory, and restart, after which they appear as downloaded. On macOS, a Bluetooth headset microphone may lower playback quality while recording, and the README suggests keeping headphones as output and selecting the built-in or an external mic instead.\n\nInstallation is not limited to one channel: the [releases page](https:\u002F\u002Fgithub.com\u002Fcjpais\u002FHandy\u002Freleases) and the [project site](https:\u002F\u002Fhandy.computer) provide builds, and Homebrew and winget packages exist but are described as not maintained by the developers. On Debian\u002FUbuntu, install the .deb with APT so dependencies resolve; using dpkg directly leads to a broken-dependency cleanup. Budget time for permissions and a first model download, and expect occasional model additions to be part of ongoing care.\n\n## HotRepo's take\n\nHotRepo has only two daily-board observations, 2026-10-24 and 2026-10-25, each showing +88 stars, and no comparable earlier history; that is not enough to judge whether interest persists. The 247-point Hacker News thread from 2026-01-15 shows discussion volume, not a quality verdict. Before committing, verify that your platform's permission flow and the model you want are covered for your hardware, especially CPU-only machines where Parakeet V3 is the relevant choice.",{"slug":25,"name":26},"productivity","Productivity & Utilities",[28,29,30,31],"accessibility","cross-platform","speech-to-text","tauri-v2","MIT","https:\u002F\u002Fhandy.computer","2026-10-25T02:46:55Z",[36,39],{"date":37,"stars":38,"stars_gained":13},"2026-10-24",34545,{"date":40,"stars":11,"stars_gained":13},"2026-10-25",1792917922]