{
  "contents": [
    {
      "role": "user",
      "parts": [
        {
          "text": "Focused auditory fact-check, no preferred ranking. Six actual sound attachments: first2 same English stop moment(full/vocal estimate); next2 same English week-six moment(full/vocal estimate); last2 English and Ukrainian complete self-worth/CTA/tail. Answer only from sound, distinguish full mix from artifact-prone Demucs voice. Q1: At 'Stop', what actually pauses: voice only, accompaniment only, both or neither? Avoid calling a short accent complete silence. Q2: At 'week six ... shock', is there really a ~2second vocal rest, or a held sung syllable/breath/brief separation? Give approximate LOCAL ranges and differentiate voiced hold from silence. Q3: In BOTH endings, locate self-worth contrast, product line, last lexical word and remaining music. Classify each phrase as melodic singing, rhythmic/narrow singing, ordinary speech, or uncertain. Do not treat narrow singing as speech automatically. Do not infer studio intent or artist identity. Foreign brand may be uncertain: write phonetic uncertainty instead of inventing wrong words or a re-recording prescription. Is the English tail silence, instrumental, humming, or mixed? Q4: How are contrastive pronouns, especially English her/mine and Ukrainian її/свою, treated by melody/duration/dynamics? Do not infer an exact octave leap or root cadence without a score. Return JSON in English<=1100words: mediaAccess(boolean),audioAccess(boolean),questions[{id,observations,evidenceLocalRanges,uncertainties}],whichClaimsNeedAFullerAnalysis,limitations. Do not fabricate exact beats/bars. These checks establish facts, not which song is10/10."
        },
        {
          "text": "Attachment 1: stop-full: targ,global101..109s;LOCAL0=global101. Actual full mix; no timing edits,constant gain only."
        },
        {
          "inlineData": {
            "mimeType": "audio/mpeg",
            "data": "[base64 of frozen bytes; see inputs.json]"
          }
        },
        {
          "text": "Attachment 2: stop-vocal: targ,global101..109s;LOCAL0=global101. Fresh Demucs vocals estimate,with leakage/separation artifacts; not a studio multitrack."
        },
        {
          "inlineData": {
            "mimeType": "audio/mpeg",
            "data": "[base64 of frozen bytes; see inputs.json]"
          }
        },
        {
          "text": "Attachment 3: six-full: targ,global133..140.3s;LOCAL0=global133. Actual full mix; no timing edits,constant gain only."
        },
        {
          "inlineData": {
            "mimeType": "audio/mpeg",
            "data": "[base64 of frozen bytes; see inputs.json]"
          }
        },
        {
          "text": "Attachment 4: six-vocal: targ,global133..140.3s;LOCAL0=global133. Fresh Demucs vocals estimate,with leakage/separation artifacts; not a studio multitrack."
        },
        {
          "inlineData": {
            "mimeType": "audio/mpeg",
            "data": "[base64 of frozen bytes; see inputs.json]"
          }
        },
        {
          "text": "Attachment 5: ending-full: targ,global176..194.375s;LOCAL0=global176. Actual full mix; no timing edits,constant gain only."
        },
        {
          "inlineData": {
            "mimeType": "audio/mpeg",
            "data": "[base64 of frozen bytes; see inputs.json]"
          }
        },
        {
          "text": "Attachment 6: ending-full: gopure,global163.4..197.726s;LOCAL0=global163.4. Actual full mix; no timing edits,constant gain only."
        },
        {
          "inlineData": {
            "mimeType": "audio/mpeg",
            "data": "[base64 of frozen bytes; see inputs.json]"
          }
        }
      ]
    }
  ],
  "stream": true,
  "generationConfig": {
    "thinkingConfig": {
      "includeThoughts": false,
      "thinkingLevel": "high"
    }
  }
}