voices.json · mesmertech/mesm-tts at main

mesm-tts / voices.json

mesm-tts: ten tiny English voices (ONNX + pdn WebAssembly engine), inference, samples, benchmarks

931612f about 23 hours ago

17.6 kB

{
 "sample_rate": 24000,
 "default_voice": "heart",
 "interface": {
  "inputs": {
   "input_ids": "int64 [1, T], misaki phoneme ids (tokenizer.json), 0 at both ends, T <= 512",
   "speed": "float32 [1]"
  },
  "outputs": {
   "waveform": "float32 [1, samples] @ 24 kHz"
  }
 },
 "level_rule": "gain_db, then y -> 0.9 + 0.1 tanh((|y| - 0.9) / 0.1) above 0.9 (soft knee)",
 "join_rule": "web/pv/flow.mjs (= tts.py join): sentences 2..n keep lead_keep_s of leading silence, sentences 1..n-1 keep gap_scale of trailing silence",
 "voices": {
  "heart": {
   "name": "Heart",
   "gender": "female",
   "accent": "us",
   "tag": "The original",
   "about": "Kokoro-82M's af_heart voice, distilled into Paradee. The fastest voice here.",
   "line": "This is where the page began: one small voice, made right here in your browser, faster than you can blink.",
   "teacher": "hexgrad/Kokoro-82M (af_heart)",
   "teacher_url": "https://huggingface.co/hexgrad/Kokoro-82M",
   "engine": "paradee",
   "gain_db": 2.4,
   "join": null,
   "default_model": "v2f_lite",
   "models": {
    "v2f_lite": {
     "onnx": "onnx/heart/v2f_lite.onnx",
     "onnx_fp32_compute": "onnx/heart/v2f_lite_noint8.onnx",
     "pdn": "web/v2/models/v2f_lite.pdn",
     "params_runtime": 3249540,
     "params_pytorch": 3309118,
     "files": [
      {
       "path": "onnx/heart/v2f_lite.onnx",
       "bytes": 2642629,
       "sha256": "35a9dd81e2ac0d52e14786a35935f6e4183dc9dd3b63b168831be2eefba5121d"
      },
      {
       "path": "onnx/heart/v2f_lite_noint8.onnx",
       "bytes": 2614959,
       "sha256": "1a80197348702930b95aa2cab8f38f5a4d43cea0449a766e58d5b59017ebe825"
      },
      {
       "path": "web/v2/models/v2f_lite.pdn",
       "bytes": 2670520,
       "sha256": "657c3ca5436768799803ea68fff11ecce88b1a4cd01df153c08d5f82c18a125a"
      }
     ],
     "test200": {
      "utmos": 4.453,
      "wer": 5.64,
      "wer_capped": 4.84
     }
    },
    "v2f": {
     "onnx": "onnx/heart/v2f.onnx",
     "onnx_fp32_compute": "onnx/heart/v2f_noint8.onnx",
     "pdn": "web/v2/models/v2f.pdn",
     "params_runtime": 4097892,
     "params_pytorch": 4157470,
     "files": [
      {
       "path": "onnx/heart/v2f.onnx",
       "bytes": 3424978,
       "sha256": "542e0439dbbeee7041c8139c0cf7ce31f01f9fb52d81cf7661cee285abd6b27e"
      },
      {
       "path": "onnx/heart/v2f_noint8.onnx",
       "bytes": 3397308,
       "sha256": "cebb3091a8e03042be56896be56c74f5f1731a5cb17f66b3092ed408ae6b0225"
      },
      {
       "path": "web/v2/models/v2f.pdn",
       "bytes": 3526456,
       "sha256": "76d9c59ce57e4c91c79747608d6c04528c97fd8c7f726ae4d2fe3cc2480010c7"
      }
     ],
     "test200": {
      "utmos": 4.462,
      "wer": 5.65,
      "wer_capped": 4.91,
      "wer_turbo": 3.37
     }
    },
    "v2g": {
     "onnx": "onnx/heart/v2g.onnx",
     "onnx_fp32_compute": "onnx/heart/v2g_noint8.onnx",
     "pdn": "web/v2/models/v2g.pdn",
     "params_runtime": 2909904,
     "params_pytorch": 2909770,
     "files": [
      {
       "path": "onnx/heart/v2g.onnx",
       "bytes": 2320277,
       "sha256": "3194a5d447e4f34333f1764d3da4ffab546acee728d57539c887c5b6b4e8eb1a"
      },
      {
       "path": "onnx/heart/v2g_noint8.onnx",
       "bytes": 2292607,
       "sha256": "067ba764ad9a35e05c6f11e70fca793d58edce221776a8a61dc18a3fd0723006"
      },
      {
       "path": "web/v2/models/v2g.pdn",
       "bytes": 2338472,
       "sha256": "e265507ca8387a9d5874748fa4b99304d0522a2964aba431150e7a5acaeb24b4"
      }
     ],
     "test200": {
      "utmos": 4.453,
      "wer": 5.48,
      "wer_capped": 4.91,
      "wer_turbo": 3.21
     }
    }
   },
   "quality": {
    "set": "TEST-200 (held out, 3 synthesis seeds)",
    "teacher_test200": {
     "utmos": 4.52,
     "wer": 5.67,
     "wer_turbo": 3.35
    }
   }
  },
  "wren": {
   "name": "Wren",
   "gender": "female",
   "accent": "us",
   "tag": "Built-in",
   "about": "Phonon's own voice: the one it speaks with when you don't choose one.",
   "line": "No server is reading this to you. Every word is being made on your device, as you listen.",
   "teacher": "gradium/phonon-31m-en (built-in voice)",
   "teacher_url": "https://huggingface.co/gradium/phonon-31m-en",
   "engine": "pv",
   "onnx": "onnx/voices/wren.onnx",
   "pdn": "web/voices/wren.pdn",
   "decoder_head": "160x4",
   "params_runtime": 5939560,
   "params_pytorch": 6019469,
   "gain_db": 1.4,
   "join": {
    "gap_scale": 0.787,
    "lead_keep_s": 0.02
   },
   "files": [
    {
     "path": "onnx/voices/wren.onnx",
     "bytes": 5340533,
     "sha256": "3b9485e8d546b437ede4c1b31d04879d0378e2167585647775e146ebf0522e93"
    },
    {
     "path": "web/voices/wren.pdn",
     "bytes": 5753308,
     "sha256": "9a48ecf23b295d7d1c6f5f59d87bcf88fa0b8d43b88d2f04bc77530e4b6b6340"
    }
   ],
   "quality": {
    "set": "DEV-400 (held out; student 3 seeds vs its own Phonon teacher 2 seeds)",
    "utmos": 4.417,
    "teacher_utmos": 4.452,
    "utmos_gap": 0.035,
    "utmos_gap_ci95": [
     0.024,
     0.047
    ],
    "wer": 6.24,
    "teacher_wer": 6.11,
    "wer_capped": 5.66,
    "teacher_wer_capped": 5.5,
    "wer_capped_diff_ci95": [
     -0.26,
     0.61
    ],
    "wer_turbo": 3.48,
    "teacher_wer_turbo": 3.61,
    "speaker_sim_to_teacher": 0.857,
    "speaker_sim_teacher_vs_itself": 0.923
   }
  },
  "everly": {
   "name": "Everly",
   "gender": "female",
   "accent": "us",
   "tag": "Soft",
   "about": "Rounded and soft-edged, with a gentle lift at the end of phrases and an easy, relaxed delivery.",
   "line": "Take a breath and settle in. I can read you anything you like, as slowly as you like.",
   "teacher": "gradium/phonon-31m-en + gradium/phonon-voices (Everly)",
   "teacher_url": "https://huggingface.co/gradium/phonon-voices",
   "engine": "pv",
   "onnx": "onnx/voices/everly.onnx",
   "pdn": "web/voices/everly.pdn",
   "decoder_head": "160x4",
   "params_runtime": 5939560,
   "params_pytorch": 6019469,
   "gain_db": 1.7,
   "join": {
    "gap_scale": 0.741,
    "lead_keep_s": 0.02
   },
   "files": [
    {
     "path": "onnx/voices/everly.onnx",
     "bytes": 5340541,
     "sha256": "5f3240766efd56333f4be0c3a52bddc8e9c3438d37707abf96ed5c8e770875ca"
    },
    {
     "path": "web/voices/everly.pdn",
     "bytes": 5753308,
     "sha256": "a3a06359a091039a0e4a18822a93b9da7819b87e8ec8ef7be30cae32ad7b05ec"
    }
   ],
   "quality": {
    "set": "DEV-400 (held out; student 3 seeds vs its own Phonon teacher 2 seeds)",
    "utmos": 4.346,
    "teacher_utmos": 4.434,
    "utmos_gap": 0.088,
    "utmos_gap_ci95": [
     0.076,
     0.1
    ],
    "wer": 6.09,
    "teacher_wer": 6.12,
    "wer_capped": 5.34,
    "teacher_wer_capped": 5.47,
    "wer_capped_diff_ci95": [
     -0.54,
     0.29
    ],
    "wer_turbo": 3.4,
    "teacher_wer_turbo": 3.91,
    "speaker_sim_to_teacher": 0.877,
    "speaker_sim_teacher_vs_itself": 0.935
   }
  },
  "hadley": {
   "name": "Hadley",
   "gender": "female",
   "accent": "us",
   "tag": "Calm",
   "about": "Settled, even and kind, with crisp consonants.",
   "line": "Your text never leaves this tab. It turns into speech right here, in a few milliseconds.",
   "teacher": "gradium/phonon-31m-en + gradium/phonon-voices (Hadley)",
   "teacher_url": "https://huggingface.co/gradium/phonon-voices",
   "engine": "pv",
   "onnx": "onnx/voices/hadley.onnx",
   "pdn": "web/voices/hadley.pdn",
   "decoder_head": "160x4",
   "params_runtime": 5939560,
   "params_pytorch": 6019469,
   "gain_db": 2.2,
   "join": {
    "gap_scale": 0.496,
    "lead_keep_s": 0.02
   },
   "files": [
    {
     "path": "onnx/voices/hadley.onnx",
     "bytes": 5340541,
     "sha256": "045069a58980ad74e141f6f61eb2a0611841dbba5a45a1fa7aac8eed17a4b2dc"
    },
    {
     "path": "web/voices/hadley.pdn",
     "bytes": 5753308,
     "sha256": "585aea65fef7f6b7482b738da9810c60424c41d2ded416e558a21911ab01b4d4"
    }
   ],
   "quality": {
    "set": "DEV-400 (held out; student 3 seeds vs its own Phonon teacher 2 seeds)",
    "utmos": 4.469,
    "teacher_utmos": 4.496,
    "utmos_gap": 0.026,
    "utmos_gap_ci95": [
     0.019,
     0.035
    ],
    "wer": 5.98,
    "teacher_wer": 6.1,
    "wer_capped": 5.52,
    "teacher_wer_capped": 5.47,
    "wer_capped_diff_ci95": [
     -0.41,
     0.51
    ],
    "wer_turbo": 3.21,
    "teacher_wer_turbo": 3.56,
    "speaker_sim_to_teacher": 0.882,
    "speaker_sim_teacher_vs_itself": 0.937
   }
  },
  "bryony": {
   "name": "Bryony",
   "gender": "female",
   "accent": "uk",
   "tag": "Warm",
   "about": "Full and rich in the mid-range, close to the mic, brisk with a little pitch play.",
   "line": "Put the kettle on, find a comfy chair, and I'll read you the next chapter.",
   "teacher": "gradium/phonon-31m-en + gradium/phonon-voices (Bryony)",
   "teacher_url": "https://huggingface.co/gradium/phonon-voices",
   "engine": "pv",
   "onnx": "onnx/voices/bryony.onnx",
   "pdn": "web/voices/bryony.pdn",
   "decoder_head": "160x4",
   "params_runtime": 5939560,
   "params_pytorch": 6019469,
   "gain_db": 0.1,
   "join": {
    "gap_scale": 0.75,
    "lead_keep_s": 0.02
   },
   "files": [
    {
     "path": "onnx/voices/bryony.onnx",
     "bytes": 5340541,
     "sha256": "80c0f9e939e64220806622bf9c86d35330fde1dd72bbdb8e535094d94f57d1ab"
    },
    {
     "path": "web/voices/bryony.pdn",
     "bytes": 5753308,
     "sha256": "78e91f5027a260912d360632f7e72f5e1121b69f3a371a3b520a48c893e25c32"
    }
   ],
   "quality": {
    "set": "DEV-400 (held out; student 3 seeds vs its own Phonon teacher 2 seeds)",
    "utmos": 4.172,
    "teacher_utmos": 4.331,
    "utmos_gap": 0.159,
    "utmos_gap_ci95": [
     0.145,
     0.173
    ],
    "wer": 6.95,
    "teacher_wer": 7.43,
    "wer_capped": 6.22,
    "teacher_wer_capped": 6.52,
    "wer_capped_diff_ci95": [
     -0.78,
     0.21
    ],
    "wer_turbo": 3.56,
    "teacher_wer_turbo": 4.17,
    "speaker_sim_to_teacher": 0.856,
    "speaker_sim_teacher_vs_itself": 0.934
   }
  },
  "maeve": {
   "name": "Maeve",
   "gender": "female",
   "accent": "uk",
   "tag": "Bright",
   "about": "Sparky and attentive; gets to the point with a smile.",
   "line": "Right, let's get straight to it. Tell me what to say, and I'll say it with a smile.",
   "teacher": "gradium/phonon-31m-en + gradium/phonon-voices (Maeve)",
   "teacher_url": "https://huggingface.co/gradium/phonon-voices",
   "engine": "pv",
   "onnx": "onnx/voices/maeve.onnx",
   "pdn": "web/voices/maeve.pdn",
   "decoder_head": "160x4",
   "params_runtime": 5939560,
   "params_pytorch": 6019469,
   "gain_db": 1.4,
   "join": {
    "gap_scale": 0.756,
    "lead_keep_s": 0.02
   },
   "files": [
    {
     "path": "onnx/voices/maeve.onnx",
     "bytes": 5340540,
     "sha256": "c690e522715d142f37f99bec07e1dbe8b7a2f2197b15a55514ee66910f2eeeb1"
    },
    {
     "path": "web/voices/maeve.pdn",
     "bytes": 5753308,
     "sha256": "ae22f81f29eef92d9908fff3b57ee1a293fd3a32de558bedc8768905005b4b90"
    }
   ],
   "quality": {
    "set": "DEV-400 (held out; student 3 seeds vs its own Phonon teacher 2 seeds)",
    "utmos": 4.391,
    "teacher_utmos": 4.442,
    "utmos_gap": 0.052,
    "utmos_gap_ci95": [
     0.042,
     0.061
    ],
    "wer": 6.31,
    "teacher_wer": 6.2,
    "wer_capped": 5.75,
    "teacher_wer_capped": 5.64,
    "wer_capped_diff_ci95": [
     -0.39,
     0.59
    ],
    "wer_turbo": 3.35,
    "teacher_wer_turbo": 3.86,
    "speaker_sim_to_teacher": 0.88,
    "speaker_sim_teacher_vs_itself": 0.94
   }
  },
  "miles": {
   "name": "Miles",
   "gender": "male",
   "accent": "us",
   "tag": "Resonant",
   "about": "Resonant and decisive, with real dynamic range: down low for a clause, then driving back up.",
   "line": "Give me a headline, and I'll make it sound like the most important news of the day.",
   "teacher": "gradium/phonon-31m-en + gradium/phonon-voices (Miles)",
   "teacher_url": "https://huggingface.co/gradium/phonon-voices",
   "engine": "pv",
   "onnx": "onnx/voices/miles.onnx",
   "pdn": "web/voices/miles.pdn",
   "decoder_head": "160x4",
   "params_runtime": 5939560,
   "params_pytorch": 6019469,
   "gain_db": 2.8,
   "join": {
    "gap_scale": 0.597,
    "lead_keep_s": 0.02
   },
   "files": [
    {
     "path": "onnx/voices/miles.onnx",
     "bytes": 5340540,
     "sha256": "ef9d59b43885b8806b100773fae52c086abc2d5afb59c4fcf3e93644856f22b6"
    },
    {
     "path": "web/voices/miles.pdn",
     "bytes": 5753308,
     "sha256": "6a7d64b949e11cc03d6997f723c6cc4c431a9d8e337efc41e0e7c796d89db54e"
    }
   ],
   "quality": {
    "set": "DEV-400 (held out; student 3 seeds vs its own Phonon teacher 2 seeds)",
    "utmos": 4.258,
    "teacher_utmos": 4.365,
    "utmos_gap": 0.107,
    "utmos_gap_ci95": [
     0.092,
     0.122
    ],
    "wer": 6.17,
    "teacher_wer": 6.45,
    "wer_capped": 5.46,
    "teacher_wer_capped": 5.53,
    "wer_capped_diff_ci95": [
     -0.49,
     0.36
    ],
    "wer_turbo": 3.26,
    "teacher_wer_turbo": 3.8,
    "speaker_sim_to_teacher": 0.88,
    "speaker_sim_teacher_vs_itself": 0.933
   }
  },
  "trevor": {
   "name": "Trevor",
   "gender": "male",
   "accent": "us",
   "tag": "Lively",
   "about": "Buoyant energy over a resonant mid-low core.",
   "line": "Hey, I'm running in your browser right now, and honestly? I'm having a great time.",
   "teacher": "gradium/phonon-31m-en + gradium/phonon-voices (Trevor)",
   "teacher_url": "https://huggingface.co/gradium/phonon-voices",
   "engine": "pv",
   "onnx": "onnx/voices/trevor.onnx",
   "pdn": "web/voices/trevor.pdn",
   "decoder_head": "160x4",
   "params_runtime": 5939560,
   "params_pytorch": 6019469,
   "gain_db": 1.2,
   "join": {
    "gap_scale": 0.905,
    "lead_keep_s": 0.02
   },
   "files": [
    {
     "path": "onnx/voices/trevor.onnx",
     "bytes": 5340541,
     "sha256": "ff8f8fba5fc1f2436e37d54c3cc7fe40499d35c5e3373dfe55e25168573cba1f"
    },
    {
     "path": "web/voices/trevor.pdn",
     "bytes": 5753308,
     "sha256": "44423ad1ebf9bbcd50fd10abc5b7718d26edecf2a7875eed0b32dfbbd4129055"
    }
   ],
   "quality": {
    "set": "DEV-400 (held out; student 3 seeds vs its own Phonon teacher 2 seeds)",
    "utmos": 4.323,
    "teacher_utmos": 4.425,
    "utmos_gap": 0.102,
    "utmos_gap_ci95": [
     0.09,
     0.114
    ],
    "wer": 6.23,
    "teacher_wer": 6.17,
    "wer_capped": 5.55,
    "teacher_wer_capped": 5.42,
    "wer_capped_diff_ci95": [
     -0.3,
     0.55
    ],
    "wer_turbo": 3.07,
    "teacher_wer_turbo": 3.61,
    "speaker_sim_to_teacher": 0.855,
    "speaker_sim_teacher_vs_itself": 0.941
   }
  },
  "edwin": {
   "name": "Edwin",
   "gender": "male",
   "accent": "uk",
   "tag": "Crisp",
   "about": "Efficient and firm, with tidy articulation.",
   "line": "Good evening. Shall we begin? I'll be narrating tonight's story, clearly and precisely.",
   "teacher": "gradium/phonon-31m-en + gradium/phonon-voices (Edwin)",
   "teacher_url": "https://huggingface.co/gradium/phonon-voices",
   "engine": "pv",
   "onnx": "onnx/voices/edwin.onnx",
   "pdn": "web/voices/edwin.pdn",
   "decoder_head": "256x6",
   "params_runtime": 7885960,
   "params_pytorch": 7965869,
   "gain_db": -1.8,
   "join": {
    "gap_scale": 0.92,
    "lead_keep_s": 0.02
   },
   "files": [
    {
     "path": "onnx/voices/edwin.onnx",
     "bytes": 6503216,
     "sha256": "4a6953592d78f0b23dda16c96a7f343d919bfc6cf84baa1cccf661b9bf25e234"
    },
    {
     "path": "web/voices/edwin.pdn",
     "bytes": 6914412,
     "sha256": "504402db4acb29f69e73247a8782b13d9ef22a881de94f53c2f6802c51f5d38c"
    }
   ],
   "quality": {
    "set": "DEV-400 (held out; student 3 seeds vs its own Phonon teacher 2 seeds)",
    "utmos": 4.216,
    "teacher_utmos": 4.389,
    "utmos_gap": 0.173,
    "utmos_gap_ci95": [
     0.153,
     0.193
    ],
    "wer": 6.05,
    "teacher_wer": 6.16,
    "wer_capped": 5.44,
    "teacher_wer_capped": 5.55,
    "wer_capped_diff_ci95": [
     -0.56,
     0.34
    ],
    "wer_turbo": 3.06,
    "teacher_wer_turbo": 3.6,
    "speaker_sim_to_teacher": 0.87,
    "speaker_sim_teacher_vs_itself": 0.938
   }
  },
  "hugh": {
   "name": "Hugh",
   "gender": "male",
   "accent": "uk",
   "tag": "Gentle",
   "about": "Soft and close, airy, with lively pitch movement and a sincere delivery.",
   "line": "Lean in a little. I'll keep it gentle, and I'll keep it short.",
   "teacher": "gradium/phonon-31m-en + gradium/phonon-voices (Hugh)",
   "teacher_url": "https://huggingface.co/gradium/phonon-voices",
   "engine": "pv",
   "onnx": "onnx/voices/hugh.onnx",
   "pdn": "web/voices/hugh.pdn",
   "decoder_head": "160x4",
   "params_runtime": 5939560,
   "params_pytorch": 6019469,
   "gain_db": -1.8,
   "join": {
    "gap_scale": 0.898,
    "lead_keep_s": 0.02
   },
   "files": [
    {
     "path": "onnx/voices/hugh.onnx",
     "bytes": 5340539,
     "sha256": "36db3b18dcdfe1e5bf7e0bd4828a0d2e7cd80556e2b8cdd03ca3026434457232"
    },
    {
     "path": "web/voices/hugh.pdn",
     "bytes": 5753308,
     "sha256": "4d6e7ec992c973ef097fbc08b838367160708f6f1a1ce0ddc31eef086f94e328"
    }
   ],
   "quality": {
    "set": "DEV-400 (held out; student 3 seeds vs its own Phonon teacher 2 seeds)",
    "utmos": 4.267,
    "teacher_utmos": 4.367,
    "utmos_gap": 0.1,
    "utmos_gap_ci95": [
     0.089,
     0.111
    ],
    "wer": 6.83,
    "teacher_wer": 6.58,
    "wer_capped": 5.94,
    "teacher_wer_capped": 5.78,
    "wer_capped_diff_ci95": [
     -0.35,
     0.64
    ],
    "wer_turbo": 3.46,
    "teacher_wer_turbo": 3.99,
    "speaker_sim_to_teacher": 0.88,
    "speaker_sim_teacher_vs_itself": 0.949
   }
  }
 }
}

	{
	"sample_rate": 24000,
	"default_voice": "heart",
	"interface": {
	"inputs": {
	"input_ids": "int64 [1, T], misaki phoneme ids (tokenizer.json), 0 at both ends, T <= 512",
	"speed": "float32 [1]"
	},
	"outputs": {
	"waveform": "float32 [1, samples] @ 24 kHz"
	}
	},
	"level_rule": "gain_db, then y -> 0.9 + 0.1 tanh((\|y\| - 0.9) / 0.1) above 0.9 (soft knee)",
	"join_rule": "web/pv/flow.mjs (= tts.py join): sentences 2..n keep lead_keep_s of leading silence, sentences 1..n-1 keep gap_scale of trailing silence",
	"voices": {
	"heart": {
	"name": "Heart",
	"gender": "female",
	"accent": "us",
	"tag": "The original",
	"about": "Kokoro-82M's af_heart voice, distilled into Paradee. The fastest voice here.",
	"line": "This is where the page began: one small voice, made right here in your browser, faster than you can blink.",
	"teacher": "hexgrad/Kokoro-82M (af_heart)",
	"teacher_url": "https://huggingface.co/hexgrad/Kokoro-82M",
	"engine": "paradee",
	"gain_db": 2.4,
	"join": null,
	"default_model": "v2f_lite",
	"models": {
	"v2f_lite": {
	"onnx": "onnx/heart/v2f_lite.onnx",
	"onnx_fp32_compute": "onnx/heart/v2f_lite_noint8.onnx",
	"pdn": "web/v2/models/v2f_lite.pdn",
	"params_runtime": 3249540,
	"params_pytorch": 3309118,
	"files": [
	{
	"path": "onnx/heart/v2f_lite.onnx",
	"bytes": 2642629,
	"sha256": "35a9dd81e2ac0d52e14786a35935f6e4183dc9dd3b63b168831be2eefba5121d"
	},
	{
	"path": "onnx/heart/v2f_lite_noint8.onnx",
	"bytes": 2614959,
	"sha256": "1a80197348702930b95aa2cab8f38f5a4d43cea0449a766e58d5b59017ebe825"
	},
	{
	"path": "web/v2/models/v2f_lite.pdn",
	"bytes": 2670520,
	"sha256": "657c3ca5436768799803ea68fff11ecce88b1a4cd01df153c08d5f82c18a125a"
	}
	],
	"test200": {
	"utmos": 4.453,
	"wer": 5.64,
	"wer_capped": 4.84
	}
	},
	"v2f": {
	"onnx": "onnx/heart/v2f.onnx",
	"onnx_fp32_compute": "onnx/heart/v2f_noint8.onnx",
	"pdn": "web/v2/models/v2f.pdn",
	"params_runtime": 4097892,
	"params_pytorch": 4157470,
	"files": [
	{
	"path": "onnx/heart/v2f.onnx",
	"bytes": 3424978,
	"sha256": "542e0439dbbeee7041c8139c0cf7ce31f01f9fb52d81cf7661cee285abd6b27e"
	},
	{
	"path": "onnx/heart/v2f_noint8.onnx",
	"bytes": 3397308,
	"sha256": "cebb3091a8e03042be56896be56c74f5f1731a5cb17f66b3092ed408ae6b0225"
	},
	{
	"path": "web/v2/models/v2f.pdn",
	"bytes": 3526456,
	"sha256": "76d9c59ce57e4c91c79747608d6c04528c97fd8c7f726ae4d2fe3cc2480010c7"
	}
	],
	"test200": {
	"utmos": 4.462,
	"wer": 5.65,
	"wer_capped": 4.91,
	"wer_turbo": 3.37
	}
	},
	"v2g": {
	"onnx": "onnx/heart/v2g.onnx",
	"onnx_fp32_compute": "onnx/heart/v2g_noint8.onnx",
	"pdn": "web/v2/models/v2g.pdn",
	"params_runtime": 2909904,
	"params_pytorch": 2909770,
	"files": [
	{
	"path": "onnx/heart/v2g.onnx",
	"bytes": 2320277,
	"sha256": "3194a5d447e4f34333f1764d3da4ffab546acee728d57539c887c5b6b4e8eb1a"
	},
	{
	"path": "onnx/heart/v2g_noint8.onnx",
	"bytes": 2292607,
	"sha256": "067ba764ad9a35e05c6f11e70fca793d58edce221776a8a61dc18a3fd0723006"
	},
	{
	"path": "web/v2/models/v2g.pdn",
	"bytes": 2338472,
	"sha256": "e265507ca8387a9d5874748fa4b99304d0522a2964aba431150e7a5acaeb24b4"
	}
	],
	"test200": {
	"utmos": 4.453,
	"wer": 5.48,
	"wer_capped": 4.91,
	"wer_turbo": 3.21
	}
	}
	},
	"quality": {
	"set": "TEST-200 (held out, 3 synthesis seeds)",
	"teacher_test200": {
	"utmos": 4.52,
	"wer": 5.67,
	"wer_turbo": 3.35
	}
	}
	},
	"wren": {
	"name": "Wren",
	"gender": "female",
	"accent": "us",
	"tag": "Built-in",
	"about": "Phonon's own voice: the one it speaks with when you don't choose one.",
	"line": "No server is reading this to you. Every word is being made on your device, as you listen.",
	"teacher": "gradium/phonon-31m-en (built-in voice)",
	"teacher_url": "https://huggingface.co/gradium/phonon-31m-en",
	"engine": "pv",
	"onnx": "onnx/voices/wren.onnx",
	"pdn": "web/voices/wren.pdn",
	"decoder_head": "160x4",
	"params_runtime": 5939560,
	"params_pytorch": 6019469,
	"gain_db": 1.4,
	"join": {
	"gap_scale": 0.787,
	"lead_keep_s": 0.02
	},
	"files": [
	{
	"path": "onnx/voices/wren.onnx",
	"bytes": 5340533,
	"sha256": "3b9485e8d546b437ede4c1b31d04879d0378e2167585647775e146ebf0522e93"
	},
	{
	"path": "web/voices/wren.pdn",
	"bytes": 5753308,
	"sha256": "9a48ecf23b295d7d1c6f5f59d87bcf88fa0b8d43b88d2f04bc77530e4b6b6340"
	}
	],
	"quality": {
	"set": "DEV-400 (held out; student 3 seeds vs its own Phonon teacher 2 seeds)",
	"utmos": 4.417,
	"teacher_utmos": 4.452,
	"utmos_gap": 0.035,
	"utmos_gap_ci95": [
	0.024,
	0.047
	],
	"wer": 6.24,
	"teacher_wer": 6.11,
	"wer_capped": 5.66,
	"teacher_wer_capped": 5.5,
	"wer_capped_diff_ci95": [
	-0.26,
	0.61
	],
	"wer_turbo": 3.48,
	"teacher_wer_turbo": 3.61,
	"speaker_sim_to_teacher": 0.857,
	"speaker_sim_teacher_vs_itself": 0.923
	}
	},
	"everly": {
	"name": "Everly",
	"gender": "female",
	"accent": "us",
	"tag": "Soft",
	"about": "Rounded and soft-edged, with a gentle lift at the end of phrases and an easy, relaxed delivery.",
	"line": "Take a breath and settle in. I can read you anything you like, as slowly as you like.",
	"teacher": "gradium/phonon-31m-en + gradium/phonon-voices (Everly)",
	"teacher_url": "https://huggingface.co/gradium/phonon-voices",
	"engine": "pv",
	"onnx": "onnx/voices/everly.onnx",
	"pdn": "web/voices/everly.pdn",
	"decoder_head": "160x4",
	"params_runtime": 5939560,
	"params_pytorch": 6019469,
	"gain_db": 1.7,
	"join": {
	"gap_scale": 0.741,
	"lead_keep_s": 0.02
	},
	"files": [
	{
	"path": "onnx/voices/everly.onnx",
	"bytes": 5340541,
	"sha256": "5f3240766efd56333f4be0c3a52bddc8e9c3438d37707abf96ed5c8e770875ca"
	},
	{
	"path": "web/voices/everly.pdn",
	"bytes": 5753308,
	"sha256": "a3a06359a091039a0e4a18822a93b9da7819b87e8ec8ef7be30cae32ad7b05ec"
	}
	],
	"quality": {
	"set": "DEV-400 (held out; student 3 seeds vs its own Phonon teacher 2 seeds)",
	"utmos": 4.346,
	"teacher_utmos": 4.434,
	"utmos_gap": 0.088,
	"utmos_gap_ci95": [
	0.076,
	0.1
	],
	"wer": 6.09,
	"teacher_wer": 6.12,
	"wer_capped": 5.34,
	"teacher_wer_capped": 5.47,
	"wer_capped_diff_ci95": [
	-0.54,
	0.29
	],
	"wer_turbo": 3.4,
	"teacher_wer_turbo": 3.91,
	"speaker_sim_to_teacher": 0.877,
	"speaker_sim_teacher_vs_itself": 0.935
	}
	},
	"hadley": {
	"name": "Hadley",
	"gender": "female",
	"accent": "us",
	"tag": "Calm",
	"about": "Settled, even and kind, with crisp consonants.",
	"line": "Your text never leaves this tab. It turns into speech right here, in a few milliseconds.",
	"teacher": "gradium/phonon-31m-en + gradium/phonon-voices (Hadley)",
	"teacher_url": "https://huggingface.co/gradium/phonon-voices",
	"engine": "pv",
	"onnx": "onnx/voices/hadley.onnx",
	"pdn": "web/voices/hadley.pdn",
	"decoder_head": "160x4",
	"params_runtime": 5939560,
	"params_pytorch": 6019469,
	"gain_db": 2.2,
	"join": {
	"gap_scale": 0.496,
	"lead_keep_s": 0.02
	},
	"files": [
	{
	"path": "onnx/voices/hadley.onnx",
	"bytes": 5340541,
	"sha256": "045069a58980ad74e141f6f61eb2a0611841dbba5a45a1fa7aac8eed17a4b2dc"
	},
	{
	"path": "web/voices/hadley.pdn",
	"bytes": 5753308,
	"sha256": "585aea65fef7f6b7482b738da9810c60424c41d2ded416e558a21911ab01b4d4"
	}
	],
	"quality": {
	"set": "DEV-400 (held out; student 3 seeds vs its own Phonon teacher 2 seeds)",
	"utmos": 4.469,
	"teacher_utmos": 4.496,
	"utmos_gap": 0.026,
	"utmos_gap_ci95": [
	0.019,
	0.035
	],
	"wer": 5.98,
	"teacher_wer": 6.1,
	"wer_capped": 5.52,
	"teacher_wer_capped": 5.47,
	"wer_capped_diff_ci95": [
	-0.41,
	0.51
	],
	"wer_turbo": 3.21,
	"teacher_wer_turbo": 3.56,
	"speaker_sim_to_teacher": 0.882,
	"speaker_sim_teacher_vs_itself": 0.937
	}
	},
	"bryony": {
	"name": "Bryony",
	"gender": "female",
	"accent": "uk",
	"tag": "Warm",
	"about": "Full and rich in the mid-range, close to the mic, brisk with a little pitch play.",
	"line": "Put the kettle on, find a comfy chair, and I'll read you the next chapter.",
	"teacher": "gradium/phonon-31m-en + gradium/phonon-voices (Bryony)",
	"teacher_url": "https://huggingface.co/gradium/phonon-voices",
	"engine": "pv",
	"onnx": "onnx/voices/bryony.onnx",
	"pdn": "web/voices/bryony.pdn",
	"decoder_head": "160x4",
	"params_runtime": 5939560,
	"params_pytorch": 6019469,
	"gain_db": 0.1,
	"join": {
	"gap_scale": 0.75,
	"lead_keep_s": 0.02
	},
	"files": [
	{
	"path": "onnx/voices/bryony.onnx",
	"bytes": 5340541,
	"sha256": "80c0f9e939e64220806622bf9c86d35330fde1dd72bbdb8e535094d94f57d1ab"
	},
	{
	"path": "web/voices/bryony.pdn",
	"bytes": 5753308,
	"sha256": "78e91f5027a260912d360632f7e72f5e1121b69f3a371a3b520a48c893e25c32"
	}
	],
	"quality": {
	"set": "DEV-400 (held out; student 3 seeds vs its own Phonon teacher 2 seeds)",
	"utmos": 4.172,
	"teacher_utmos": 4.331,
	"utmos_gap": 0.159,
	"utmos_gap_ci95": [
	0.145,
	0.173
	],
	"wer": 6.95,
	"teacher_wer": 7.43,
	"wer_capped": 6.22,
	"teacher_wer_capped": 6.52,
	"wer_capped_diff_ci95": [
	-0.78,
	0.21
	],
	"wer_turbo": 3.56,
	"teacher_wer_turbo": 4.17,
	"speaker_sim_to_teacher": 0.856,
	"speaker_sim_teacher_vs_itself": 0.934
	}
	},
	"maeve": {
	"name": "Maeve",
	"gender": "female",
	"accent": "uk",
	"tag": "Bright",
	"about": "Sparky and attentive; gets to the point with a smile.",
	"line": "Right, let's get straight to it. Tell me what to say, and I'll say it with a smile.",
	"teacher": "gradium/phonon-31m-en + gradium/phonon-voices (Maeve)",
	"teacher_url": "https://huggingface.co/gradium/phonon-voices",
	"engine": "pv",
	"onnx": "onnx/voices/maeve.onnx",
	"pdn": "web/voices/maeve.pdn",
	"decoder_head": "160x4",
	"params_runtime": 5939560,
	"params_pytorch": 6019469,
	"gain_db": 1.4,
	"join": {
	"gap_scale": 0.756,
	"lead_keep_s": 0.02
	},
	"files": [
	{
	"path": "onnx/voices/maeve.onnx",
	"bytes": 5340540,
	"sha256": "c690e522715d142f37f99bec07e1dbe8b7a2f2197b15a55514ee66910f2eeeb1"
	},
	{
	"path": "web/voices/maeve.pdn",
	"bytes": 5753308,
	"sha256": "ae22f81f29eef92d9908fff3b57ee1a293fd3a32de558bedc8768905005b4b90"
	}
	],
	"quality": {
	"set": "DEV-400 (held out; student 3 seeds vs its own Phonon teacher 2 seeds)",
	"utmos": 4.391,
	"teacher_utmos": 4.442,
	"utmos_gap": 0.052,
	"utmos_gap_ci95": [
	0.042,
	0.061
	],
	"wer": 6.31,
	"teacher_wer": 6.2,
	"wer_capped": 5.75,
	"teacher_wer_capped": 5.64,
	"wer_capped_diff_ci95": [
	-0.39,
	0.59
	],
	"wer_turbo": 3.35,
	"teacher_wer_turbo": 3.86,
	"speaker_sim_to_teacher": 0.88,
	"speaker_sim_teacher_vs_itself": 0.94
	}
	},
	"miles": {
	"name": "Miles",
	"gender": "male",
	"accent": "us",
	"tag": "Resonant",
	"about": "Resonant and decisive, with real dynamic range: down low for a clause, then driving back up.",
	"line": "Give me a headline, and I'll make it sound like the most important news of the day.",
	"teacher": "gradium/phonon-31m-en + gradium/phonon-voices (Miles)",
	"teacher_url": "https://huggingface.co/gradium/phonon-voices",
	"engine": "pv",
	"onnx": "onnx/voices/miles.onnx",
	"pdn": "web/voices/miles.pdn",
	"decoder_head": "160x4",
	"params_runtime": 5939560,
	"params_pytorch": 6019469,
	"gain_db": 2.8,
	"join": {
	"gap_scale": 0.597,
	"lead_keep_s": 0.02
	},
	"files": [
	{
	"path": "onnx/voices/miles.onnx",
	"bytes": 5340540,
	"sha256": "ef9d59b43885b8806b100773fae52c086abc2d5afb59c4fcf3e93644856f22b6"
	},
	{
	"path": "web/voices/miles.pdn",
	"bytes": 5753308,
	"sha256": "6a7d64b949e11cc03d6997f723c6cc4c431a9d8e337efc41e0e7c796d89db54e"
	}
	],
	"quality": {
	"set": "DEV-400 (held out; student 3 seeds vs its own Phonon teacher 2 seeds)",
	"utmos": 4.258,
	"teacher_utmos": 4.365,
	"utmos_gap": 0.107,
	"utmos_gap_ci95": [
	0.092,
	0.122
	],
	"wer": 6.17,
	"teacher_wer": 6.45,
	"wer_capped": 5.46,
	"teacher_wer_capped": 5.53,
	"wer_capped_diff_ci95": [
	-0.49,
	0.36
	],
	"wer_turbo": 3.26,
	"teacher_wer_turbo": 3.8,
	"speaker_sim_to_teacher": 0.88,
	"speaker_sim_teacher_vs_itself": 0.933
	}
	},
	"trevor": {
	"name": "Trevor",
	"gender": "male",
	"accent": "us",
	"tag": "Lively",
	"about": "Buoyant energy over a resonant mid-low core.",
	"line": "Hey, I'm running in your browser right now, and honestly? I'm having a great time.",
	"teacher": "gradium/phonon-31m-en + gradium/phonon-voices (Trevor)",
	"teacher_url": "https://huggingface.co/gradium/phonon-voices",
	"engine": "pv",
	"onnx": "onnx/voices/trevor.onnx",
	"pdn": "web/voices/trevor.pdn",
	"decoder_head": "160x4",
	"params_runtime": 5939560,
	"params_pytorch": 6019469,
	"gain_db": 1.2,
	"join": {
	"gap_scale": 0.905,
	"lead_keep_s": 0.02
	},
	"files": [
	{
	"path": "onnx/voices/trevor.onnx",
	"bytes": 5340541,
	"sha256": "ff8f8fba5fc1f2436e37d54c3cc7fe40499d35c5e3373dfe55e25168573cba1f"
	},
	{
	"path": "web/voices/trevor.pdn",
	"bytes": 5753308,
	"sha256": "44423ad1ebf9bbcd50fd10abc5b7718d26edecf2a7875eed0b32dfbbd4129055"
	}
	],
	"quality": {
	"set": "DEV-400 (held out; student 3 seeds vs its own Phonon teacher 2 seeds)",
	"utmos": 4.323,
	"teacher_utmos": 4.425,
	"utmos_gap": 0.102,
	"utmos_gap_ci95": [
	0.09,
	0.114
	],
	"wer": 6.23,
	"teacher_wer": 6.17,
	"wer_capped": 5.55,
	"teacher_wer_capped": 5.42,
	"wer_capped_diff_ci95": [
	-0.3,
	0.55
	],
	"wer_turbo": 3.07,
	"teacher_wer_turbo": 3.61,
	"speaker_sim_to_teacher": 0.855,
	"speaker_sim_teacher_vs_itself": 0.941
	}
	},
	"edwin": {
	"name": "Edwin",
	"gender": "male",
	"accent": "uk",
	"tag": "Crisp",
	"about": "Efficient and firm, with tidy articulation.",
	"line": "Good evening. Shall we begin? I'll be narrating tonight's story, clearly and precisely.",
	"teacher": "gradium/phonon-31m-en + gradium/phonon-voices (Edwin)",
	"teacher_url": "https://huggingface.co/gradium/phonon-voices",
	"engine": "pv",
	"onnx": "onnx/voices/edwin.onnx",
	"pdn": "web/voices/edwin.pdn",
	"decoder_head": "256x6",
	"params_runtime": 7885960,
	"params_pytorch": 7965869,
	"gain_db": -1.8,
	"join": {
	"gap_scale": 0.92,
	"lead_keep_s": 0.02
	},
	"files": [
	{
	"path": "onnx/voices/edwin.onnx",
	"bytes": 6503216,
	"sha256": "4a6953592d78f0b23dda16c96a7f343d919bfc6cf84baa1cccf661b9bf25e234"
	},
	{
	"path": "web/voices/edwin.pdn",
	"bytes": 6914412,
	"sha256": "504402db4acb29f69e73247a8782b13d9ef22a881de94f53c2f6802c51f5d38c"
	}
	],
	"quality": {
	"set": "DEV-400 (held out; student 3 seeds vs its own Phonon teacher 2 seeds)",
	"utmos": 4.216,
	"teacher_utmos": 4.389,
	"utmos_gap": 0.173,
	"utmos_gap_ci95": [
	0.153,
	0.193
	],
	"wer": 6.05,
	"teacher_wer": 6.16,
	"wer_capped": 5.44,
	"teacher_wer_capped": 5.55,
	"wer_capped_diff_ci95": [
	-0.56,
	0.34
	],
	"wer_turbo": 3.06,
	"teacher_wer_turbo": 3.6,
	"speaker_sim_to_teacher": 0.87,
	"speaker_sim_teacher_vs_itself": 0.938
	}
	},
	"hugh": {
	"name": "Hugh",
	"gender": "male",
	"accent": "uk",
	"tag": "Gentle",
	"about": "Soft and close, airy, with lively pitch movement and a sincere delivery.",
	"line": "Lean in a little. I'll keep it gentle, and I'll keep it short.",
	"teacher": "gradium/phonon-31m-en + gradium/phonon-voices (Hugh)",
	"teacher_url": "https://huggingface.co/gradium/phonon-voices",
	"engine": "pv",
	"onnx": "onnx/voices/hugh.onnx",
	"pdn": "web/voices/hugh.pdn",
	"decoder_head": "160x4",
	"params_runtime": 5939560,
	"params_pytorch": 6019469,
	"gain_db": -1.8,
	"join": {
	"gap_scale": 0.898,
	"lead_keep_s": 0.02
	},
	"files": [
	{
	"path": "onnx/voices/hugh.onnx",
	"bytes": 5340539,
	"sha256": "36db3b18dcdfe1e5bf7e0bd4828a0d2e7cd80556e2b8cdd03ca3026434457232"
	},
	{
	"path": "web/voices/hugh.pdn",
	"bytes": 5753308,
	"sha256": "4d6e7ec992c973ef097fbc08b838367160708f6f1a1ce0ddc31eef086f94e328"
	}
	],
	"quality": {
	"set": "DEV-400 (held out; student 3 seeds vs its own Phonon teacher 2 seeds)",
	"utmos": 4.267,
	"teacher_utmos": 4.367,
	"utmos_gap": 0.1,
	"utmos_gap_ci95": [
	0.089,
	0.111
	],
	"wer": 6.83,
	"teacher_wer": 6.58,
	"wer_capped": 5.94,
	"teacher_wer_capped": 5.78,
	"wer_capped_diff_ci95": [
	-0.35,
	0.64
	],
	"wer_turbo": 3.46,
	"teacher_wer_turbo": 3.99,
	"speaker_sim_to_teacher": 0.88,
	"speaker_sim_teacher_vs_itself": 0.949
	}
	}
	}
	}