v150 · local-first · speech workflow

Local Triage Workflow

A concrete app pattern for local speech: dictate sensitive notes, classify the transcript in-page, and generate follow-up fields while the privacy state stays visible.

Dictation has not started. This page requests local processing before listening.

SpeechRecognition checking

Local processing control checking

Language pack not checked

Network checking

Privacy processLocally requires audio and transcript processing on this device

triage output


      
speech event log

      
No speech events yet.

    

  


  

    
recognition setup

    
const Recognition = window.SpeechRecognition || window.webkitSpeechRecognition;
const recognition = new Recognition();
recognition.lang = "en-US";
recognition.continuous = true;
recognition.interimResults = true;
recognition.processLocally = true;

const packStatus = await Recognition.available({
  langs: ["en-US"],
  processLocally: true,
});

if (packStatus === "available") {
  recognition.start();
} else if (packStatus === "downloadable") {
  await Recognition.install({ langs: ["en-US"], processLocally: true });
}

  


  

    
see also