Show HN: A tiny browser-based voice-to-text tool (no signup, no tracking) (www.voicetotextonline.com)

A lightweight, no‑signup browser voice‑to‑text tool was posted on Show HN that promises instant, privacy‑focused transcription of speech into text in 30+ languages (Hindi, English, Spanish, Arabic, French, etc.). The app runs entirely in the browser using the Web Speech API, claims to process audio locally so nothing is stored or sent to its servers, and offers real‑time dictation for students, writers, and professionals without downloads or accounts. It includes language‑specific guides, device tips, and recommends Chrome/Edge for best support (Safari partial; Firefox unsupported). For the AI/ML community the project is a practical example of democratizing speech recognition via browser APIs: low friction, zero‑infrastructure deployment and improved privacy/UX for everyday use. Technically, it leverages client‑side browser speech recognition rather than shipping a custom ASR model or server pipeline, so accuracy and capabilities depend on the browser, OS speech engine, and microphone quality. That makes it great for prototyping, accessible dictation, and privacy‑first workflows, but not a replacement for server‑side ASR when you need batch processing, advanced diarization, or bespoke model tuning.