{
  "schema": "meccanoid-audio-qa-v1",
  "created_at": "2026-09-21T14:48:41.299Z",
  "passed": 9,
  "tests": [
    "cached local ASR and local speech present",
    "local text-to-speech creates WAV and two measured images",
    "reordered identical request replays without speaking again",
    "same request id with different text rejected",
    "real speech waveform transcribed by cached Whisper; no typed-text shortcut",
    "silent PCM produces no hallucinated transcript",
    "malformed WAV rejected",
    "audio observation preserves transcript/source and two image attachments",
    "MCP result contains native image and audio blocks"
  ],
  "model": "openai/whisper-tiny.en",
  "model_revision": "87c7102498dcde7456f24cfd30239ca606ed9063",
  "spoken_text": "Hello robot. This is a local robot simulation. Please move to the green marker.",
  "actual_transcript": "The Lowscot, this in a local robot simulation. Please move to the green marker.",
  "word_error_rate": 0.21428571428571427,
  "duration_seconds": 5.4706875,
  "transcription_seconds": 2.1066667712293565,
  "local_inference": true,
  "source": "Synthetic espeak-ng speech fixture, not a real microphone",
  "microphone_hardware_tested": false,
  "privacy": "Only this explicitly synthetic fixture is public. User-supplied WAV and future microphone audio remain private.",
  "waveform": "audio/synthetic-demo-waveform.png",
  "spectrogram": "audio/synthetic-demo-spectrogram.png",
  "audio": "audio/synthetic-demo.wav"
}