{
  "schema_version": 1,
  "episode_1_active_models": [
    "qwen3_tts",
    "voxcpm2",
    "higgs_tts_3_4b",
    "supertonic3",
    "audio8_preview_0.6b"
  ],
  "episode_1_excluded_models": {
    "moss_tts_local_v1.5": "청취 후 제작자 판단으로 1편 비교군에서 제외. 비감정 원본만 보존한다."
  },
  "models": {
    "qwen3_tts": {
      "display_name": "Qwen3-TTS",
      "filename_prefix": "qwen3_tts",
      "emotion_control_supported": true,
      "method": "generate_custom_voice(..., instruct=자연어 지시)",
      "official_local_source": "vendors/qwen3-tts/README.md#custom-voice-generate",
      "emotion_prompts": {
        "ko_pacing": "Speak in Korean at a natural conversational pace with only short, natural pauses between sentences. Do not speak slowly or stretch syllables. Finish the full passage in about 18 to 22 seconds.",
        "ko_emotion_neutral": "Speak this sentence in a calm, neutral voice.",
        "ko_emotion_happy": "Very happy, excited, joyful, and energetic.",
        "ko_emotion_angry": "Speak with an extremely angry emotion. Shout with intense anger.",
        "ko_emotion_whisper": "Speak only in a very quiet, breathy whisper. Do not use a normal speaking voice."
      }
    },
    "voxcpm2": {
      "display_name": "VoxCPM2",
      "filename_prefix": "voxcpm2",
      "emotion_control_supported": true,
      "method": "합성문 앞에 (voice/control description)을 붙이는 공식 Voice Design 형식",
      "official_local_source": "vendors/voxcpm/README.md#-voice-design",
      "emotion_prompts": {
        "ko_emotion_neutral": "calm and neutral",
        "ko_emotion_happy": "extremely joyful, excited, bright and energetic",
        "ko_emotion_angry": "furious, shouting loudly, fast and intense",
        "ko_emotion_whisper": "breathy whisper, extremely quiet, close to the microphone"
      }
    },
    "higgs_tts_3_4b": {
      "display_name": "Higgs TTS 3",
      "filename_prefix": "higgs_tts_3",
      "emotion_control_supported": true,
      "method": "문장 시작에 Higgs 전용 <|category:value|> 토큰 배치",
      "official_local_source": "vendors/sglang-omni/docs/cookbook/higgs_tts.md#inline-control-tokens",
      "emotion_prompts": {
        "ko_emotion_neutral": "<|prosody:expressive_low|>",
        "ko_emotion_happy": "<|emotion:elation|><|prosody:expressive_high|><|prosody:pitch_high|>",
        "ko_emotion_angry": "<|emotion:anger|><|style:shouting|><|prosody:expressive_high|>",
        "ko_emotion_whisper": "<|style:whispering|>"
      }
    },
    "supertonic3": {
      "display_name": "Supertonic 3",
      "filename_prefix": "supertonic3",
      "emotion_control_supported": false,
      "method": "고정 voice-style과 speed, 10개 표현 태그를 제공하지만 angry/happy 같은 일반 감정 지시는 제공하지 않음",
      "official_local_source": "vendors/supertonic/README.md"
    },
    "audio8_preview_0.6b": {
      "display_name": "Audio8 Preview 0.6B",
      "filename_prefix": "audio8",
      "emotion_control_supported": false,
      "method": "text와 선택적 reference_audio/reference_text만 제공; 감정 또는 style 지시 인자 없음",
      "official_local_source": "vendors/audio8/README.md#inference",
      "episode_1_condition": "남녀 비교를 위해 동일한 합성 레퍼런스를 사용한 reference-conditioned TTS"
    }
  }
}