{
  "model_path": "models/Qwen3-TTS-12Hz-0.6B-Base",
  "reference_audio": "examples/readme_clone_input.wav",
  "reference_text": "Okay. Yeah. I resent you. I love you. I respect you. But you know what? You blew it! And thanks to you.",
  "synthesis_text": "Hello.",
  "language": "English",
  "dtype": "float32",
  "device": "cpu",
  "x_vector_only_mode": true,
  "non_streaming_mode": false,
  "generation": {
    "max_new_tokens": 64,
    "do_sample": false,
    "subtalker_dosample": false,
    "top_k": null,
    "top_p": null,
    "temperature": null,
    "subtalker_top_k": null,
    "subtalker_top_p": null,
    "subtalker_temperature": null
  },
  "token_ids": {
    "tts_bos_token_id": 151672,
    "tts_eos_token_id": 151673,
    "tts_pad_token_id": 151671,
    "codec_bos_id": 2149,
    "codec_eos_token_id": 2150,
    "codec_pad_id": 2148,
    "codec_think_id": 2154,
    "codec_think_bos_id": 2156,
    "codec_think_eos_id": 2157,
    "language_id": 2050
  },
  "shapes": {
    "input_ids": [
      1,
      10
    ],
    "speaker_embedding": [
      1024
    ],
    "prefill_embedding": [
      1,
      10,
      1024
    ],
    "trailing_text_hidden": [
      1,
      2,
      1024
    ],
    "tts_pad_embed": [
      1,
      1,
      1024
    ],
    "speech_codes": [
      63,
      16
    ],
    "hidden_states": [
      63,
      1024
    ],
    "first_frame_logits": [
      1,
      3072
    ],
    "decoded_audio": [
      120960
    ]
  },
  "outputs": {
    "det_text_tokens.bin": {
      "path": "det_text_tokens.bin",
      "shape": [
        1,
        10
      ],
      "dtype": "int64",
      "size_bytes": 80,
      "sha256": "e474aaf2b4f77ae954b12bd11a4df3421c10002c3a769015c14bf15ec4818d69"
    },
    "det_speaker_embedding.bin": {
      "path": "det_speaker_embedding.bin",
      "shape": [
        1024
      ],
      "dtype": "float32",
      "size_bytes": 4096,
      "sha256": "ceb9b83df983f0d6febd9694efba7c54baaaabab2fb8c269f7f4540e55dde62d"
    },
    "det_prefill_embedding.bin": {
      "path": "det_prefill_embedding.bin",
      "shape": [
        1,
        10,
        1024
      ],
      "dtype": "float32",
      "size_bytes": 40960,
      "sha256": "508c08984e9aec5240996c3b7c909816a257d77054a313e5da8358d071434195"
    },
    "det_speech_codes.bin": {
      "path": "det_speech_codes.bin",
      "shape": [
        63,
        16
      ],
      "dtype": "int64",
      "size_bytes": 8064,
      "sha256": "f0a9f60c562c2d928335307bde7b5350a39032f6d3c4b942db3b1b36f7d81d9e"
    },
    "det_hidden_states.bin": {
      "path": "det_hidden_states.bin",
      "shape": [
        63,
        1024
      ],
      "dtype": "float32",
      "size_bytes": 258048,
      "sha256": "4f18c5b447d8df052113b2f1506afed84d93e81703db1e84ad1b6500345b73e0"
    },
    "det_first_frame_logits.bin": {
      "path": "det_first_frame_logits.bin",
      "shape": [
        1,
        3072
      ],
      "dtype": "float32",
      "size_bytes": 12288,
      "sha256": "ee73ee94eb8d9f6dafa1659cd103c863555b1a9d49242dd8df84076de1925a74"
    },
    "det_decoded_audio.bin": {
      "path": "det_decoded_audio.bin",
      "shape": [
        120960
      ],
      "dtype": "float32",
      "size_bytes": 483840,
      "sha256": "9a68b3cacbfce9aa4a4c83242d923ed16061c0bead5245d5cd21c4d4ef576d7c"
    },
    "det_trailing_text.bin": {
      "path": "det_trailing_text.bin",
      "shape": [
        1,
        2,
        1024
      ],
      "dtype": "float32",
      "size_bytes": 8192,
      "sha256": "1f4b9d63069fe9e984b49d45b73536f56701be90a1a560fed411ffe205671a28"
    },
    "det_tts_pad_embed.bin": {
      "path": "det_tts_pad_embed.bin",
      "shape": [
        1,
        1,
        1024
      ],
      "dtype": "float32",
      "size_bytes": 4096,
      "sha256": "5601144c3b8cf4875be3bcf03aa4ffe31a52466e1eb5bec01a30776eaec09a8e"
    },
    "det_role_embed.bin": {
      "path": "det_role_embed.bin",
      "shape": [
        1,
        3,
        1024
      ],
      "dtype": "float32",
      "size_bytes": 12288,
      "sha256": "d0a63bdf6318fafbf8d4d967cbdd4624d3018edde348e73ae4e55676e6b78833"
    },
    "det_codec_prefill_embed.bin": {
      "path": "det_codec_prefill_embed.bin",
      "shape": [
        1,
        6,
        1024
      ],
      "dtype": "float32",
      "size_bytes": 24576,
      "sha256": "43fe88411e4f350e87482cff865393a35adb99efc0517adf5f28b3227a84e78a"
    },
    "det_tts_overlay.bin": {
      "path": "det_tts_overlay.bin",
      "shape": [
        1,
        6,
        1024
      ],
      "dtype": "float32",
      "size_bytes": 24576,
      "sha256": "f69534f5378ea69190c10156d6c67c3a9f1b03e5bd99d04682cce49a8e13b6bb"
    }
  },
  "wav": {
    "path": "det_output.wav",
    "sha256": "c42960e886ac35f312dcfdd1d1c42fdf492e083f0c05b4ceb4575dcc808616b7",
    "sample_rate": 24000,
    "num_samples": 120960,
    "duration_seconds": 5.04
  },
  "prefill_token_types": "det_prefill_token_types.json"
}