Spaces:
Running
Running
File size: 21,859 Bytes
9361148 2f01cc6 9361148 2f01cc6 9361148 86371bb 2f01cc6 86371bb 9361148 86371bb 2f01cc6 86371bb 2f01cc6 86371bb 2f01cc6 86371bb 2f01cc6 86371bb 2f01cc6 86371bb 9361148 2f01cc6 9361148 2f01cc6 9361148 86371bb 2f01cc6 86371bb 9361148 86371bb 9361148 86371bb 9361148 2f01cc6 9361148 2f01cc6 9361148 86371bb 9361148 38e8ba6 86371bb 38e8ba6 9361148 38e8ba6 86371bb 38e8ba6 86371bb 38e8ba6 9361148 38e8ba6 86371bb 38e8ba6 86371bb 38e8ba6 86371bb 38e8ba6 9361148 38e8ba6 9361148 38e8ba6 9361148 38e8ba6 9361148 38e8ba6 9361148 86371bb 38e8ba6 9361148 38e8ba6 9361148 86371bb 38e8ba6 9361148 38e8ba6 86371bb 38e8ba6 86371bb 38e8ba6 9361148 38e8ba6 86371bb 38e8ba6 86371bb 38e8ba6 86371bb 38e8ba6 9361148 2f01cc6 9361148 2f01cc6 9361148 2f01cc6 9361148 2f01cc6 9361148 2f01cc6 9361148 2f01cc6 9361148 2f01cc6 9361148 2f01cc6 9361148 2f01cc6 9361148 | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 | # -*- coding: utf-8 -*-
"""
简单单字导出插件
从TextGrid提取分词片段,按拼音排序导出
"""
import os
import json
import glob
import shutil
import logging
from typing import Any, Dict, List, Tuple
from .base import ExportPlugin, PluginOption, OptionType
logger = logging.getLogger(__name__)
class SimpleExportPlugin(ExportPlugin):
"""简单单字导出插件"""
name = "简单单字导出"
description = "从TextGrid提取分词片段,按时长排序导出"
version = "1.1.0"
author = "内置"
def get_options(self) -> List[PluginOption]:
return [
PluginOption(
key="max_samples",
label="每个拼音最大样本数",
option_type=OptionType.NUMBER,
default=10,
min_value=1,
max_value=1000,
description="按质量评分排序,保留最佳的N个"
),
PluginOption(
key="quality_metrics",
label="质量评估维度",
option_type=OptionType.COMBO,
default="duration",
choices=["duration", "duration+rms", "duration+f0", "all"],
description="duration=仅时长, +rms=音量稳定性, +f0=音高稳定性。选择 all 可能耗时较长"
),
PluginOption(
key="extend_duration",
label="头尾拓展(秒)",
option_type=OptionType.TEXT,
default="0",
description="裁剪时头尾各拓展的时长,最大0.5秒。若一边到达边界,另一边继续拓展"
),
PluginOption(
key="naming_rule",
label="命名规则",
option_type=OptionType.TEXT,
default="%p%%n%",
description="变量: %p%=拼音, %n%=序号。示例: %p%_%n% → ba_1.wav"
),
PluginOption(
key="first_naming_rule",
label="首个样本命名规则",
option_type=OptionType.TEXT,
default="%p%",
description="第0个样本的特殊规则,留空则使用通用规则。示例: %p% → ba.wav"
),
PluginOption(
key="clean_temp",
label="导出后清理临时文件",
option_type=OptionType.SWITCH,
default=True,
description="删除临时的segments目录"
)
]
def _apply_extend(
self,
start_time: float,
end_time: float,
extend_duration: float,
audio_duration: float
) -> Tuple[float, float]:
"""
应用头尾拓展
头尾各拓展 extend_duration 秒,若一边到达边界则另一边继续拓展
"""
if extend_duration <= 0:
return start_time, end_time
total_extend = extend_duration * 2
# 先尝试两边各拓展
new_start = max(0, start_time - extend_duration)
new_end = min(audio_duration, end_time + extend_duration)
# 计算实际拓展量,剩余量补偿到另一边
used = (start_time - new_start) + (new_end - end_time)
remaining = total_extend - used
if remaining > 0:
# 优先补偿到尾部,再补偿到头部
extra_end = min(remaining, audio_duration - new_end)
new_end += extra_end
remaining -= extra_end
if remaining > 0:
new_start = max(0, new_start - remaining)
return new_start, new_end
def export(
self,
source_name: str,
bank_dir: str,
options: Dict[str, Any]
) -> Tuple[bool, str]:
"""执行简单单字导出"""
try:
# 使用基类方法获取语言设置
language = self.load_language_from_meta(bank_dir, source_name)
max_samples = int(options.get("max_samples", 10))
naming_rule = options.get("naming_rule", "%p%_%n%")
first_naming_rule = options.get("first_naming_rule", "")
clean_temp = options.get("clean_temp", True)
quality_metrics = options.get("quality_metrics", "duration")
# 使用基类方法解析质量评估维度
enabled_metrics = self.parse_quality_metrics(quality_metrics)
paths = self.get_source_paths(bank_dir, source_name)
export_dir = self.get_export_dir(bank_dir, source_name, "simple_export")
# 临时segments目录
temp_base = os.path.join(bank_dir, ".temp_segments")
segments_dir = os.path.join(temp_base, source_name)
# 获取头尾拓展参数
extend_duration = min(float(options.get("extend_duration", 0)), 0.5)
# 步骤1: 提取分词片段
self._log("【提取分词片段】")
if extend_duration > 0:
self._log(f"头尾拓展: {extend_duration}s(单边到达边界时另一边继续拓展)")
success, msg, pinyin_counts = self._extract_segments(
paths["slices_dir"],
paths["textgrid_dir"],
segments_dir,
language,
extend_duration
)
if not success:
return False, msg
# 步骤2: 排序导出
self._log(f"\n【排序导出】评估维度: {enabled_metrics}")
success, msg = self._sort_and_export(
segments_dir,
export_dir,
max_samples,
naming_rule,
first_naming_rule,
enabled_metrics
)
if not success:
return False, msg
# 清理临时目录
if clean_temp and os.path.exists(segments_dir):
self._log(f"\n清理临时目录: {segments_dir}")
shutil.rmtree(segments_dir)
if os.path.exists(temp_base) and not os.listdir(temp_base):
shutil.rmtree(temp_base)
return True, f"导出完成: {export_dir}"
except Exception as e:
logger.error(f"简单单字导出失败: {e}", exc_info=True)
return False, str(e)
def _extract_segments(
self,
slices_dir: str,
textgrid_dir: str,
segments_dir: str,
language: str,
extend_duration: float = 0.0
) -> Tuple[bool, str, Dict[str, int]]:
"""
提取分词片段
中文:使用words层按字切分,用char_to_pinyin获取拼音名称
日语:使用phones层按音素切分,合并辅音+元音为音节
参数:
extend_duration: 头尾拓展总时长(秒),单边到达边界时另一边继续拓展
"""
try:
import textgrid
import soundfile as sf
os.makedirs(segments_dir, exist_ok=True)
tg_files = glob.glob(os.path.join(textgrid_dir, '*.TextGrid'))
if not tg_files:
return False, "未找到TextGrid文件", {}
self._log(f"处理 {len(tg_files)} 个TextGrid文件")
# 根据语言选择提取方法
if language in ("japanese", "ja", "jp"):
return self._extract_japanese_segments(
tg_files, slices_dir, segments_dir, extend_duration
)
else:
return self._extract_chinese_segments(
tg_files, slices_dir, segments_dir, language, extend_duration
)
except Exception as e:
logger.error(f"提取分词失败: {e}", exc_info=True)
return False, str(e), {}
def _extract_chinese_segments(
self,
tg_files: List[str],
slices_dir: str,
segments_dir: str,
language: str,
extend_duration: float = 0.0
) -> Tuple[bool, str, Dict[str, int]]:
"""
中文音频提取
使用words层的时间边界,按字符切分,用char_to_pinyin获取拼音
参数:
extend_duration: 头尾拓展总时长(秒),单边到达边界时另一边继续拓展
"""
import textgrid
import soundfile as sf
from src.text_processor import char_to_pinyin, is_valid_char
pinyin_counts: Dict[str, int] = {}
for tg_path in tg_files:
basename = os.path.basename(tg_path).replace('.TextGrid', '.wav')
wav_path = os.path.join(slices_dir, basename)
if not os.path.exists(wav_path):
self._log(f"警告: 找不到 {basename}")
continue
tg = textgrid.TextGrid.fromFile(tg_path)
audio, sr = sf.read(wav_path, dtype='float32')
audio_duration = len(audio) / sr
# 使用words层(第一层)
words_tier = tg[0]
for interval in words_tier:
word_text = interval.mark.strip()
if not word_text or word_text in ['', 'SP', 'AP', '<unk>', 'spn', 'sil']:
continue
start_time = interval.minTime
end_time = interval.maxTime
duration = end_time - start_time
# 获取有效字符
chars = list(word_text)
valid_chars = [c for c in chars if is_valid_char(c, language)]
if not valid_chars:
continue
# 按字符均分时长
char_duration = duration / len(valid_chars)
for i, char in enumerate(valid_chars):
pinyin = char_to_pinyin(char, language)
if not pinyin:
continue
char_start = start_time + i * char_duration
char_end = char_start + char_duration
# 应用头尾拓展,单边到达边界时另一边继续拓展
actual_start, actual_end = self._apply_extend(
char_start, char_end, extend_duration, audio_duration
)
pinyin_dir = os.path.join(segments_dir, pinyin)
os.makedirs(pinyin_dir, exist_ok=True)
current_count = pinyin_counts.get(pinyin, 0)
index = current_count + 1
pinyin_counts[pinyin] = index
start_sample = int(round(actual_start * sr))
end_sample = int(round(actual_end * sr))
segment = audio[start_sample:end_sample]
if len(segment) == 0:
pinyin_counts[pinyin] = current_count
continue
output_path = os.path.join(pinyin_dir, f'{index}.wav')
sf.write(output_path, segment, sr, subtype='PCM_16')
total = sum(pinyin_counts.values())
self._log(f"提取完成: {len(pinyin_counts)} 个拼音,共 {total} 个片段")
return True, f"提取完成: {len(pinyin_counts)} 个拼音", pinyin_counts
def _extract_japanese_segments(
self,
tg_files: List[str],
slices_dir: str,
segments_dir: str,
extend_duration: float = 0.0
) -> Tuple[bool, str, Dict[str, int]]:
"""
日语音频提取
使用phones层,将辅音+元音合并为音节
参数:
extend_duration: 头尾拓展总时长(秒),单边到达边界时另一边继续拓展
"""
import textgrid
import soundfile as sf
phone_counts: Dict[str, int] = {}
for tg_path in tg_files:
basename = os.path.basename(tg_path).replace('.TextGrid', '.wav')
wav_path = os.path.join(slices_dir, basename)
if not os.path.exists(wav_path):
self._log(f"警告: 找不到 {basename}")
continue
tg = textgrid.TextGrid.fromFile(tg_path)
audio, sr = sf.read(wav_path, dtype='float32')
audio_duration = len(audio) / sr
# 查找phones层
phones_tier = None
for tier in tg:
if tier.name.lower() in ('phones', 'phone'):
phones_tier = tier
break
if phones_tier is None and len(tg) >= 2:
phones_tier = tg[1]
if phones_tier is None:
self._log(f"警告: {basename} 未找到phones层,跳过")
continue
# 合并音素为音节
syllables = self._merge_japanese_phones(phones_tier)
for syllable, start_time, end_time in syllables:
if not syllable:
continue
# 标准化为ASCII
normalized = self._normalize_japanese_phone(syllable)
if not normalized:
continue
# 应用头尾拓展,单边到达边界时另一边继续拓展
actual_start, actual_end = self._apply_extend(
start_time, end_time, extend_duration, audio_duration
)
phone_dir = os.path.join(segments_dir, normalized)
os.makedirs(phone_dir, exist_ok=True)
current_count = phone_counts.get(normalized, 0)
index = current_count + 1
phone_counts[normalized] = index
start_sample = int(round(actual_start * sr))
end_sample = int(round(actual_end * sr))
segment = audio[start_sample:end_sample]
if len(segment) == 0:
phone_counts[normalized] = current_count
continue
output_path = os.path.join(phone_dir, f'{index}.wav')
sf.write(output_path, segment, sr, subtype='PCM_16')
total = sum(phone_counts.values())
self._log(f"提取完成: {len(phone_counts)} 个音节,共 {total} 个片段")
return True, f"提取完成: {len(phone_counts)} 个音节", phone_counts
def _merge_japanese_phones(self, phones_tier) -> List[Tuple[str, float, float]]:
"""
日语音素合并
规则:辅音 + 元音 合并为一个音节
"""
# 元音集合
vowels = {'a', 'e', 'i', 'o', 'u', 'ɯ'}
skip_marks = {'', 'SP', 'AP', '<unk>', 'spn', 'sil'}
syllables = []
pending_consonant = None
pending_start = None
for interval in phones_tier:
phone = interval.mark.strip()
if phone in skip_marks:
pending_consonant = None
pending_start = None
continue
# 移除长音符号判断元音
base_phone = phone.rstrip('ː')
is_vowel = base_phone in vowels
if is_vowel:
if pending_consonant is not None:
syllable = pending_consonant + phone
syllables.append((syllable, pending_start, interval.maxTime))
pending_consonant = None
pending_start = None
else:
syllables.append((phone, interval.minTime, interval.maxTime))
else:
if pending_consonant is not None:
syllables.append((pending_consonant, pending_start, interval.minTime))
pending_consonant = phone
pending_start = interval.minTime
if pending_consonant is not None:
syllables.append((pending_consonant, pending_start, phones_tier[-1].maxTime))
return syllables
def _normalize_japanese_phone(self, phone: str) -> str:
"""
日语音素标准化为ASCII
"""
# IPA到罗马音的映射
ipa_map = {
# 元音
'ɯ': 'u',
'ɯː': 'u',
'aː': 'a',
'eː': 'e',
'iː': 'i',
'oː': 'o',
'uː': 'u',
# 辅音
'ɲ': 'n',
'ŋ': 'n',
'ɕ': 'sh',
'ʑ': 'j',
'dʑ': 'j',
'tɕ': 'ch',
'ɡ': 'g',
'ː': '',
}
result = phone
# 按长度降序处理映射
for ipa in sorted(ipa_map.keys(), key=len, reverse=True):
if ipa in result:
result = result.replace(ipa, ipa_map[ipa])
# 移除非ASCII字符
result = ''.join(c for c in result if c.isascii() and c.isalnum())
return result.lower() if result else None
def _sort_and_export(
self,
segments_dir: str,
export_dir: str,
max_samples: int,
naming_rule: str,
first_naming_rule: str,
enabled_metrics: List[str]
) -> Tuple[bool, str]:
"""排序并导出"""
try:
import soundfile as sf
from src.quality_scorer import QualityScorer, duration_score
os.makedirs(export_dir, exist_ok=True)
# 清空已有导出
for f in os.listdir(export_dir):
fp = os.path.join(export_dir, f)
if os.path.isfile(fp):
os.remove(fp)
wav_files = glob.glob(
os.path.join(segments_dir, '**', '*.wav'),
recursive=True
)
if not wav_files:
return False, "未找到分字片段"
self._log(f"扫描到 {len(wav_files)} 个片段")
# 判断是否需要加载音频计算质量分数
need_audio_scoring = any(m in enabled_metrics for m in ["rms", "f0"])
# 按拼音分组
stats: Dict[str, List[Tuple[str, float, float]]] = {} # pinyin -> [(path, duration, score)]
if need_audio_scoring:
scorer = QualityScorer(enabled_metrics=enabled_metrics)
for path in wav_files:
rel_path = os.path.relpath(path, segments_dir)
parts = rel_path.split(os.sep)
if len(parts) >= 2:
pinyin = parts[0]
if pinyin not in stats:
stats[pinyin] = []
try:
info = sf.info(path)
duration = info.duration
if need_audio_scoring:
# 加载音频计算质量分数
audio, sr = sf.read(path)
if len(audio.shape) > 1:
audio = audio.mean(axis=1)
scores = scorer.score(audio, sr, duration)
quality_score = scores.get("combined", 0.5)
else:
# 仅使用时长评分
quality_score = duration_score(duration)
stats[pinyin].append((path, duration, quality_score))
except Exception as e:
logger.warning(f"处理文件失败 {path}: {e}")
continue
self._log(f"统计到 {len(stats)} 个拼音")
self._log(f"命名规则: {naming_rule}")
if first_naming_rule:
self._log(f"首个样本规则: {first_naming_rule}")
# 按质量分数排序并导出
exported = 0
for pinyin, files in stats.items():
sorted_files = sorted(files, key=lambda x: -x[2]) # 按质量分数降序
for idx, (src_path, _, _) in enumerate(sorted_files[:max_samples]):
# 使用基类方法应用命名规则
if idx == 0 and first_naming_rule:
filename = self.apply_naming_rule(first_naming_rule, pinyin, idx)
else:
filename = self.apply_naming_rule(naming_rule, pinyin, idx)
dst_path = os.path.join(export_dir, f'{filename}.wav')
shutil.copyfile(src_path, dst_path)
exported += 1
self._log(f"导出完成: {exported} 个文件")
return True, f"导出完成: {len(stats)} 个拼音,{exported} 个文件"
except Exception as e:
logger.error(f"排序导出失败: {e}", exc_info=True)
return False, str(e)
|