What it does Artyom.js is a TypeScript-to-JavaScript wrapper around webkitSpeechRecognition and speechSynthesis. It lets you register voice commands (including wildcards and regexes), dictate text, and make the browser talk back. Ships as both a CommonJS module and a plain <script> tag file.

The interesting bit The library includes a Soundex algorithm to fuzzy-match misheard commands, and it can chunk through “extreme huge blocks of text” — the README claims +20K words — while still firing onStart and onEnd callbacks. The real story, though, is the giant Chrome 71 breaking change plastered at the top of the README: Google now requires a user gesture before any speech synthesis will fire, so your “Jarvis” needs a click-to-activate button before it can run hands-free.

Key highlights

• Smart commands with wildcards (Repeat after me *) and regex support
• Optional wake-word mode (say “Jarvis” before a command)
• Supports ~13 languages including English variants, Spanish, German, Japanese, Mandarin, Cantonese
• Includes a local HTTPS sandbox at localhost:3000 for testing continuous recognition
• Single-file TypeScript source; build scripts for both browser and module targets

Caveats

• Chrome 71+ requires at least one user click per page before speech synthesis works; the README calls this out as a hard limitation, not a bug
• The “complete” language support claim is unclear — the table cuts off and some flag labels look suspect (“Brazil” uses pt-PT, not pt-BR)
• Last meaningful update appears to center on the Chrome 71 workaround; activity level is unclear from the README alone

Verdict Worth a look if you need quick voice prototyping in a controlled web app where you can guarantee a click interaction. Skip it if you need fully autonomous voice startup or cross-browser support beyond Chrome/Edge’s Web Speech implementations.