web/voices/chinese-xiaoya/meta.json
| 1 | { |
| 2 | "key": "chinese-xiaoya", |
| 3 | "espeak_voice": null, |
| 4 | "g2p_voice_slot": null, |
| 5 | "length_scale": 1.0, |
| 6 | "sample_rate": 22050, |
| 7 | "front_bytes": 1647694, |
| 8 | "dec_bytes": 1461858, |
| 9 | "front_dims": { |
| 10 | "d_vocab": 73, |
| 11 | "d_hidden": 64, |
| 12 | "d_depth": 3, |
| 13 | "d_kernel": 5, |
| 14 | "d_max_tokens": 432, |
| 15 | "d_max_duration": 80, |
| 16 | "a_vocab": 73, |
| 17 | "a_hidden": 96, |
| 18 | "a_token_depth": 3, |
| 19 | "a_depth": 4, |
| 20 | "a_kernel": 5, |
| 21 | "a_out_channels": 192, |
| 22 | "adapter_mode": 0, |
| 23 | "adapter_kernel": 0, |
| 24 | "adapter_rank": 0, |
| 25 | "weight_floats": 823563, |
| 26 | "meta_bytes": 568, |
| 27 | "n_tensors": 62 |
| 28 | }, |
| 29 | "dec_dims": { |
| 30 | "in_ch": 192, |
| 31 | "c0": 160, |
| 32 | "c1": 80, |
| 33 | "c2": 40, |
| 34 | "c3": 20, |
| 35 | "pf_channels": 0, |
| 36 | "pf_layers": 0, |
| 37 | "pf_kernel": 9, |
| 38 | "pf_scale": 0.25, |
| 39 | "hop": 256, |
| 40 | "weight_floats": 730721, |
| 41 | "meta_bytes": 416, |
| 42 | "n_tensors": 46 |
| 43 | }, |
| 44 | "checkpoints": { |
| 45 | "duration": "artifacts/voices/zh_CN/duration/duration-student.pt", |
| 46 | "acoustic": "artifacts/voices/zh_CN/joint/latent-student.pt", |
| 47 | "decoder": "artifacts/voices/zh_CN/joint/decoder-student.pt" |
| 48 | }, |
| 49 | "phoneme_type": "pinyin", |
| 50 | "phoneme_id_map": { |
| 51 | "_": 0, |
| 52 | "^": 1, |
| 53 | "$": 2, |
| 54 | "\u00d8": 3, |
| 55 | "b": 4, |
| 56 | "p": 5, |
| 57 | "m": 6, |
| 58 | "f": 7, |
| 59 | "d": 8, |
| 60 | "t": 9, |
| 61 | "n": 10, |
| 62 | "l": 11, |
| 63 | "g": 12, |
| 64 | "k": 13, |
| 65 | "h": 14, |
| 66 | "j": 15, |
| 67 | "q": 16, |
| 68 | "x": 17, |
| 69 | "zh": 18, |
| 70 | "ch": 19, |
| 71 | "sh": 20, |
| 72 | "r": 21, |
| 73 | "z": 22, |
| 74 | "c": 23, |
| 75 | "s": 24, |
| 76 | "y": 25, |
| 77 | "w": 26, |
| 78 | "a": 27, |
| 79 | "o": 28, |
| 80 | "e": 29, |
| 81 | "ai": 30, |
| 82 | "ei": 31, |
| 83 | "ao": 32, |
| 84 | "ou": 33, |
| 85 | "an": 34, |
| 86 | "en": 35, |
| 87 | "ang": 36, |
| 88 | "eng": 37, |
| 89 | "ong": 38, |
| 90 | "i": 39, |
| 91 | "ia": 40, |
| 92 | "ie": 41, |
| 93 | "iao": 42, |
| 94 | "iu": 43, |
| 95 | "ian": 44, |
| 96 | "in": 45, |
| 97 | "iang": 46, |
| 98 | "ing": 47, |
| 99 | "iong": 48, |
| 100 | "u": 49, |
| 101 | "ua": 50, |
| 102 | "uo": 51, |
| 103 | "uai": 52, |
| 104 | "ui": 53, |
| 105 | "uan": 54, |
| 106 | "un": 55, |
| 107 | "uang": 56, |
| 108 | "ueng": 57, |
| 109 | "v": 58, |
| 110 | "ve": 59, |
| 111 | "van": 60, |
| 112 | "vn": 61, |
| 113 | "er": 62, |
| 114 | "ue": 63, |
| 115 | "1": 64, |
| 116 | "2": 65, |
| 117 | "3": 66, |
| 118 | "4": 67, |
| 119 | "5": 68, |
| 120 | "\u3002": 69, |
| 121 | ".": 69, |
| 122 | "\uff1f": 70, |
| 123 | "?": 70, |
| 124 | "\uff01": 71, |
| 125 | "!": 71, |
| 126 | "\u2014": 72, |
| 127 | "\u2026": 72, |
| 128 | "\u3001": 72, |
| 129 | "\uff0c": 72, |
| 130 | ",": 72, |
| 131 | "\uff1a": 72, |
| 132 | ":": 72, |
| 133 | "\uff1b": 72, |
| 134 | ";": 72, |
| 135 | " ": 72 |
| 136 | }, |
| 137 | "phoneme_id_map_source": "pypkg/tests/data/zh-xiaoya-piper-phoneme-config.json", |
| 138 | "front": "front_f16.bin", |
| 139 | "front_widened_sha256": "8a0727f2ad7a4bb40a7e1e8facce73da7961a128341199c1b48e975e14cc60d9", |
| 140 | "dec": "dec_f16.bin", |
| 141 | "dec_widened_sha256": "311563e942831fc353ca4923255e5175bfde786b5e5fee5cd776cffa7aab8c2a", |
| 142 | "weights": "f16", |
| 143 | "weights_note": "float16 storage, widened to float32 by the loader before it reaches the runtime. float16 is the reference for this voice: it is the precision of the roota.raw-fp16.v1 package every CER/WER number was measured on." |
| 144 | } |
| 145 | |