{
  "id": "labzero-ai/zero-talk-11m",
  "name": "Zero Talk 11M",
  "version": "1",
  "task": "text-to-speech",
  "summary": "An English text-to-speech model with about 11M parameters. It runs in the browser on the user's own device; nothing typed leaves the device.",
  "maker": { "name": "Lab Zero", "url": "https://labzero.ai/" },
  "page": "https://labzero.ai/zero-talk-11m/",
  "languages": ["en"],
  "voices": 1,
  "parameters": { "approximate": "11M", "detail": "10.8M plus about 94k for the voice" },
  "audio": { "sample_rate_hz": 24000, "channels": 1, "download_format": "Ogg Opus (the demo's .ogg link)" },
  "runs_on": {
    "browser": {
      "needs": ["WebAssembly", "Web Workers"],
      "engines": ["WebGPU, where the browser offers it on a secure page", "WebAssembly, everywhere else"],
      "network": "The files are fetched once, when the demo starts; speech is made on the device."
    }
  },
  "tested": [
    "Chromium on an Apple-silicon laptop, both engines",
    "Safari 26 on an iPhone 17 Pro, both engines"
  ],
  "measurements": [
    {
      "what": "speed in the browser",
      "device": "iPhone 17 Pro, Safari 26",
      "result": "8.0 s of speech made in about 3.2 s, about 2.5 times faster than real time",
      "runs": 1,
      "date": "2026-10-29",
      "build": "an earlier build of this page"
    }
  ],
  "limitations": [
    "English only, one voice.",
    "Questions come out flat; exclamations do not come across.",
    "Some S sounds have a slight lisp.",
    "Long sentences can lose their rhythm; sentences that open with a greeting and a comma can sound unnatural.",
    "Very short utterances can sound slightly cut off at the start.",
    "No controls for tempo or pauses.",
    "Accented Latin words are read in plain spelling; other scripts are left out with a warning.",
    "Some names are read oddly.",
    "The two browser engines give close but not identical audio."
  ],
  "not_available": {
    "reserved_for_upcoming_paid_version": ["control over expressive delivery styles and emphasis", "a second voice"],
    "not_shipped_yet": ["the command-line program", "natural intonation for questions", "confirmed support in Firefox, Safari on macOS, Android and mid-range phones"],
    "not_possible_with_this_model": ["languages other than English", "voice cloning"]
  },
  "builds_on": [{ "name": "Kokoro-82M", "license": "Apache-2.0", "author": "hexgrad" }],
  "license": {
    "spdx": "Apache-2.0",
    "text": "https://labzero.ai/zero-talk-11m/LICENSE.txt",
    "notice": "https://labzero.ai/zero-talk-11m/NOTICE.txt",
    "third_party_notices": "https://labzero.ai/zero-talk-11m/THIRD-PARTY-NOTICES.txt"
  },
  "files": []
}