{
  "schema_version": 2,
  "created_at": "2026-08-22T15:10:34Z",
  "claim": "A local MiniMax Music 3 run produced a 20-second synth-pop track with an intelligible three-line English vocal passage.",
  "runtime": {
    "mere_run_version": "0.43.0",
    "executed_source_commit": "76b5f90dc24f026644b74a33d56e882181f837cf",
    "merged_main_commit": "33204fb69da3b3c3117d93850b3e192b8830e9a1",
    "source_tree": "17535d68a4cf1e2b4e89e4272a88700e31779c58",
    "merge_pull_request": 357,
    "model_id": "music-minimax-music3",
    "source_repository": "MiniMaxAI/MiniMax-Music3",
    "source_revision": "bd348f9c49ea3c1b39f33ace3436f8fad435f24e",
    "performance_mode": "optimized",
    "compute_precision": "bf16",
    "sampling_tier": "quality",
    "inference_steps": 30,
    "flow_strategy": "sequential",
    "loading_strategy": "staged",
    "seed_strategy": "legacy",
    "seed": 2121,
    "guidance_scale": 1.7
  },
  "conditioning": {
    "caption": "Global Metadata\nBasic Attributes: cinematic maritime synth-pop, 104 BPM, D major. Warm, luminous, compact twenty-second song with a clear verse-to-chorus lift.\nVocal Details\nVocal Gender and Timbre: warm female lead, natural tone, close-miked and front-center. Vocal Style: precise syllabic English diction, no ad-libs, no melisma, no backing vocals, minimal reverb for maximum lyric clarity.\nArrangement\nPrimary: analog bass, brushed drums, glassy synth chords, and a restrained maritime atmosphere. The verse stays sparse; the chorus widens while leaving the lead vocal unobscured.",
    "supplied_lyrics": "[verse]\nHarbor lights are drawing silver lines\nNight rolls in, the signal still shines\n\n[chorus]\nKeep the whole world close tonight\nBeneath the northern lights",
    "published_passage": "Harbor lights are drawing silver lines. Night rolls in, the signal still shines. Keep the whole world close tonight.",
    "omitted_supplied_line": "Beneath the northern lights",
    "omission_reason": "The 500-frame, 20-second semantic budget ended after the first chorus line. Neither recognizer detected the final supplied line, so it is not part of the showcase claim."
  },
  "generation": {
    "generated_frame_count": 500,
    "native_sample_rate": 44100,
    "channels": 2,
    "source_duration_seconds": 20.015601,
    "source_size_bytes": 5296172,
    "source_wav_sha256": "bb5933d20d605b5a2b6c9ad6751ead9c250b15186138d64cae8bc9501a6c20f6",
    "source_recipe_sha256": "88459776cd3025a12863233c1e118f3963c828d9b134c068a036b5c2b01a8b0d",
    "profile_sha256": "08819ab4c697bf4bd9001ebf013378df1371b9d757c117004bdb9378f4ba99d0",
    "profile": {
      "total_seconds": 552.084760584,
      "semantic_sampling_seconds": 8.339189667,
      "flow_transformer_seconds": 164.106130332,
      "vocoder_decode_seconds": 3.167917751,
      "chunk_count": 4
    },
    "audio_health_before_export": {
      "peak": 0.92578125,
      "rms": 0.12472319,
      "sample_count": 882688,
      "stereo_collapse_fraction": 0.0
    }
  },
  "publication": {
    "reason": "Publish the complete 20-second passage and apply a 250 ms terminal fade. The conditioning section identifies the supplied line that does not appear in the clip.",
    "trim_start_seconds": 0.0,
    "trim_end_seconds": 20.0,
    "fade_out_start_seconds": 19.75,
    "fade_out_duration_seconds": 0.25,
    "wav": {
      "path": "music-minimax3-harbor.wav",
      "codec": "pcm_s24le",
      "sample_rate": 44100,
      "channels": 2,
      "duration_seconds": 20.0,
      "size_bytes": 5292102,
      "sha256": "aed830593426802a056106de47915d77b8b2a752db8018a16f671529173475d6"
    },
    "browser_audio": {
      "path": "music-minimax3-harbor.m4a",
      "codec": "aac",
      "sample_rate": 44100,
      "channels": 2,
      "duration_seconds": 20.0,
      "size_bytes": 663083,
      "sha256": "ae6252aa918426fd69e747fbb748377f51b98c5fa495e3ffe8ef4ad8954d7f13"
    },
    "measured_mean_volume_db": -18.4,
    "measured_peak_volume_db": -1.0
  },
  "vocal_validation": {
    "separation_model": {
      "id": "music-separate-bs-roformer-viperx-1297",
      "repository": "AEmotionStudio/roformer-models",
      "revision": "d323194290f8488ea51814143806609bfbd7a1e5",
      "weights_sha256": "fa296577206144929917601636b65ccdc407b6e6c2f209e4312d9d2b7b975a8a"
    },
    "source_vocal_stem_sha256": "5d5033d34e58eb0f6ad4b050fcd52bf44846c2b0da243239ad4cab8f01efd699",
    "expected_published_text": "Harbor lights are drawing silver lines. Night rolls in, the signal still shines. Keep the whole world close tonight.",
    "browser_audio_parakeet_transcript": "Harbor lights are drawing silver lines, night rolls in the signals to shines. Keep the whole world close tonight.",
    "browser_audio_qwen3_asr_transcript": "Harbor lights are drawing silver lines. Night rolls in, the signal still shines. Keep the whole world close tonight.",
    "qwen3_asr_five_second_windows": [
      "Harbor lights are drawing silhouettes.",
      "For light, night rolls.",
      "The signal still shines.",
      "Keep the whole world close tonight."
    ],
    "interpretation": "Qwen3-ASR matches the published passage verbatim. Parakeet makes one grammatical substitution in the middle line. These transcripts assess intelligibility, not phoneme-level alignment."
  },
  "diagnostic_boundary": "This result covers one prompt, seed, and 20-second generation. The clip includes three of the four supplied lyric lines. Other prompts, seeds, and durations can produce different vocal clarity and lyric coverage."
}