demo ยท v139

Availability & Language Matrix

Probe the current on-device Web Speech hooks for language readiness. The lab calls SpeechRecognition.available() with processLocally and quality, installs packs with SpeechRecognition.install(), and then starts a recognizer with processLocally = true.

SpeechRecognitionchecking

processLocallychecking

available()checking

install()checking

Permissions Policychecking

Networkchecking

Probe controls

processLocally = true Audio and transcript are required to stay on this device. Missing packs return downloadable, downloading, or a language-not-supported error when recognition starts.
processLocally = false The user agent can use either local or service-based recognition. This may work when no local language pack is installed, but the page cannot promise local privacy.

Language matrix

Language Local pack status Remote-allowed status Actions

Live result

Choose a language, then probe availability. Recognition results and install outcomes appear here.

    Current API pattern

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

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

if (status === "downloadable" || status === "downloading") {
  await Recognition.install({
    langs: ["en-US"],
    processLocally: true,
    quality: "dictation",
  });
}

recognition.onresult = (event) => {
  render(event.results[0][0].transcript);
};
recognition.start();

    see also