File size: 7,881 Bytes
c8a0bb3 5e74c24 c8a0bb3 5e74c24 c8a0bb3 5e74c24 c8a0bb3 5e74c24 c8a0bb3 5e74c24 c8a0bb3 5e74c24 c8a0bb3 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 |
#!/usr/bin/env python3
import hashlib
import json
import unittest
from dataclasses import dataclass
from pathlib import Path
_DIR = Path(__file__).parent
_REPO_DIR = _DIR.parent
@dataclass
class Language:
native: str
english: str
country: str
_LANGUAGES = {
"ca_ES": Language("Català", "Catalan", "Spain"),
"da_DK": Language("Dansk", "Danish", "Denmark"),
"de_DE": Language("Deutsch", "German", "Germany"),
"el_GR": Language("Ελληνικά", "Greek", "Greece"),
"en_GB": Language("English", "English", "Great Britain"),
"en_US": Language("English", "English", "United States"),
"es_ES": Language("Español", "Spanish", "Spain"),
"es_MX": Language("Español", "Spanish", "Mexico"),
"fi_FI": Language("Suomi", "Finnish", "Finland"),
"fr_FR": Language("Français", "French", "France"),
"is_IS": Language("íslenska", "Icelandic", "Iceland"),
"it_IT": Language("Italiano", "Italian", "Italy"),
"ka_GE": Language("ქართული ენა", "Georgian", "Georgia"),
"kk_KZ": Language("қазақша", "Kazakh", "Kazakhstan"),
"ne_NP": Language("नेपाली", "Nepali", "Nepal"),
"nl_BE": Language("Nederlands", "Dutch", "Belgium"),
"nl_NL": Language("Nederlands", "Dutch", "Netherlands"),
"no_NO": Language("Norsk", "Norwegian", "Norway"),
"pl_PL": Language("Polski", "Polish", "Poland"),
"pt_BR": Language("Português", "Portuguese", "Brazil"),
"ru_RU": Language("Русский", "Russian", "Russia"),
"sv_SE": Language("Svenska", "Swedish", "Sweden"),
"sw_CD": Language("Kiswahili", "Swahili", "Democratic Republic of the Congo"),
"uk_UA": Language("украї́нська мо́ва", "Ukrainian", "Ukraine"),
"vi_VN": Language("Tiếng Việt", "Vietnamese", "Vietnam"),
"zh_CN": Language("简体中文", "Chinese", "China"),
}
class VoiceTest(unittest.TestCase):
def test_voices(self):
for onnx_path in _REPO_DIR.rglob("*.onnx"):
with self.subTest(onnx_path=onnx_path):
self.assertGreater(onnx_path.stat().st_size, 0, "Empty onnx file")
# Load JSON config for voice
config_path = onnx_path.parent / f"{onnx_path.name}.json"
with open(config_path, "r", encoding="utf-8") as config_file:
config = json.load(config_file)
# Verify config
self.assertIn(
"piper_version", config, "Missing piper_version in config"
)
self.assertIn("language", config, "Missing language in config")
self.assertIn("dataset", config, "Missing dataset in config")
self.assertIn(
"quality", config["audio"], "Missing audio quality in config"
)
# Verify directory structure
# <lang_family>/<lang_code>/<dataset>/<quality>/
quality_dir = onnx_path.parent
dataset_dir = quality_dir.parent
lang_code_dir = dataset_dir.parent
lang_family_dir = lang_code_dir.parent
self.assertEqual(
lang_family_dir.name,
config["language"]["family"],
"Wrong lang family dir",
)
self.assertEqual(
lang_code_dir.name,
config["language"]["code"],
"Wrong lang code dir",
)
self.assertEqual(
dataset_dir.name, config["dataset"], "Wrong dataset dir"
)
self.assertEqual(
quality_dir.name, config["audio"]["quality"], "Wrong quality dir"
)
self.assertIn(lang_code_dir.name, _LANGUAGES, "Unknown language code")
# Verify file names
file_lang_code, file_dataset, file_quality = onnx_path.stem.split("-")
file_lang_family = file_lang_code.split("_", maxsplit=1)[0]
self.assertEqual(
file_lang_family,
config["language"]["family"],
"Wrong lang family file",
)
self.assertEqual(
file_lang_code, config["language"]["code"], "Wrong lang code file"
)
self.assertEqual(file_dataset, config["dataset"], "Wrong dataset file")
self.assertEqual(
file_quality, config["audio"]["quality"], "Wrong quality file"
)
def run_tests():
runner = unittest.TextTestRunner()
runner.run(unittest.makeSuite(VoiceTest))
# -----------------------------------------------------------------------------
def write_voices_json():
# {
# "<family>_<region>-<dataset>-<quality>": {
# "name": "<dataset>",
# "language": {
# "code": "<family>_<region>",
# "family": "<family>",
# "region": "<region>",
# "name_native": "<native>",
# "name_english": "<english>",
# "country_english": "<country>",
# },
# "quality": "<quality>", // x_low, low, medium, high
# "num_speakers": int,
# "speaker_id_map": {
# "name": int,
# ...
# }
# "files": {
# "relative/path/to/file": {
# "size_bytes": int,
# "md5_digest": str, // hex
# },
# ...
# }
# },
# ...
# }
voices = {}
for onnx_path in _REPO_DIR.rglob("*.onnx"):
voice_dir = onnx_path.parent
config_path = voice_dir / f"{onnx_path.name}.json"
with open(config_path, "r", encoding="utf-8") as config_file:
config = json.load(config_file)
quality = config["audio"]["quality"]
dataset = config["dataset"]
lang_code = config["language"]["code"]
lang_family, lang_region = lang_code.split("_", maxsplit=1)
lang_names = _LANGUAGES[lang_code]
voice_key = f"{lang_code}-{dataset}-{quality}"
model_card_path = voice_dir / "MODEL_CARD"
assert model_card_path.exists(), f"Missing {model_card_path}"
voices[voice_key] = {
"name": dataset,
"language": {
"code": lang_code,
"family": lang_family,
"region": lang_region,
"name_native": lang_names.native,
"name_english": lang_names.english,
"country_english": lang_names.country,
},
"quality": quality,
"num_speakers": config["num_speakers"],
"speaker_id_map": config.get("speaker_id_map", {}),
"files": {
str(file_path.relative_to(_REPO_DIR)): {
"size_bytes": file_path.stat().st_size,
"md5_digest": get_file_hash(file_path),
}
for file_path in (
onnx_path,
config_path,
model_card_path,
)
},
}
with open(_REPO_DIR / "voices.json", "w", encoding="utf-8") as voices_file:
json.dump(voices, voices_file, indent=4, ensure_ascii=False)
def get_file_hash(path, bytes_per_chunk: int = 8192) -> str:
"""Hash a file in chunks using md5."""
path_hash = hashlib.md5()
with open(path, "rb") as path_file:
chunk = path_file.read(bytes_per_chunk)
while chunk:
path_hash.update(chunk)
chunk = path_file.read(bytes_per_chunk)
return path_hash.hexdigest()
# -----------------------------------------------------------------------------
if __name__ == "__main__":
run_tests()
print("Writing voices.json")
write_voices_json()
|