""" ASR 引擎语言标识符映射 统一使用全称小写英文作为标准键（如 auto, chinese, english, japanese）， 各引擎通过映射表转换为自身所需的标识符格式。 引擎标识符格式差异: - SenseVoice: 短代码 (auto/zh/en/ja/ko/yue) - Qwen3-ASR / ForceAligner: 英文明称首字母大写 (Chinese/English/Japanese/...) - FunASR-Nano: 中文文本 (中文/英文/日文/...)，非 MLT 版仅官方验证 3 语言 - Paraformer: 不支持语言选择（中文专用模型） """ from typing import Dict, Optional, List # ── 统一语言代码列表（按使用频率排序）── # key 为统一代码（全称小写），各引擎子字典为该引擎的标识符 LANGUAGE_MAP: Dict[str, Dict[str, Optional[str]]] = { "auto": { # 自动检测 "paraformer": None, "sensevoice": "auto", "fun_asr_nano": None, "qwen_asr": None, "aligner": None, }, "chinese": { # 中文 "paraformer": None, "sensevoice": "zh", "fun_asr_nano": "中文", "qwen_asr": "Chinese", "aligner": "Chinese", }, "english": { # 英文 "paraformer": None, "sensevoice": "en", "fun_asr_nano": "英文", "qwen_asr": "English", "aligner": "English", }, "cantonese": { # 粤语 "paraformer": None, "sensevoice": "yue", "fun_asr_nano": None, "qwen_asr": "Cantonese", "aligner": "Cantonese", }, "japanese": { # 日语 "paraformer": None, "sensevoice": "ja", "fun_asr_nano": "日文", "qwen_asr": "Japanese", "aligner": "Japanese", }, "korean": { # 韩语 "paraformer": None, "sensevoice": "ko", "fun_asr_nano": None, "qwen_asr": "Korean", "aligner": "Korean", }, # ── Qwen3 额外支持的语言 ── # 注: FunASR 仅支持 中文/英文/日文，MLT 版支持更多语言 "arabic": { "paraformer": None, "sensevoice": None, "fun_asr_nano": None, "qwen_asr": "Arabic", "aligner": "Arabic", }, "german": { "paraformer": None, "sensevoice": None, "fun_asr_nano": None, "qwen_asr": "German", "aligner": "German", }, "french": { "paraformer": None, "sensevoice": None, "fun_asr_nano": None, "qwen_asr": "French", "aligner": "French", }, "spanish": { "paraformer": None, "sensevoice": None, "fun_asr_nano": None, "qwen_asr": "Spanish", "aligner": "Spanish", }, "portuguese": { "paraformer": None, "sensevoice": None, "fun_asr_nano": None, "qwen_asr": "Portuguese", "aligner": "Portuguese", }, "indonesian": { "paraformer": None, "sensevoice": None, "fun_asr_nano": None, "qwen_asr": "Indonesian", "aligner": "Indonesian", }, "italian": { "paraformer": None, "sensevoice": None, "fun_asr_nano": None, "qwen_asr": "Italian", "aligner": "Italian", }, "russian": { "paraformer": None, "sensevoice": None, "fun_asr_nano": None, "qwen_asr": "Russian", "aligner": "Russian", }, "thai": { "paraformer": None, "sensevoice": None, "fun_asr_nano": None, "qwen_asr": "Thai", "aligner": "Thai", }, "vietnamese": { "paraformer": None, "sensevoice": None, "fun_asr_nano": None, "qwen_asr": "Vietnamese", "aligner": "Vietnamese", }, "turkish": { "paraformer": None, "sensevoice": None, "fun_asr_nano": None, "qwen_asr": "Turkish", "aligner": "Turkish", }, "hindi": { "paraformer": None, "sensevoice": None, "fun_asr_nano": None, "qwen_asr": "Hindi", "aligner": "Hindi", }, "malay": { "paraformer": None, "sensevoice": None, "fun_asr_nano": None, "qwen_asr": "Malay", "aligner": "Malay", }, "dutch": { "paraformer": None, "sensevoice": None, "fun_asr_nano": None, "qwen_asr": "Dutch", "aligner": "Dutch", }, "swedish": { "paraformer": None, "sensevoice": None, "fun_asr_nano": None, "qwen_asr": "Swedish", "aligner": "Swedish", }, "danish": { "paraformer": None, "sensevoice": None, "fun_asr_nano": None, "qwen_asr": "Danish", "aligner": "Danish", }, "finnish": { "paraformer": None, "sensevoice": None, "fun_asr_nano": None, "qwen_asr": "Finnish", "aligner": "Finnish", }, "polish": { "paraformer": None, "sensevoice": None, "fun_asr_nano": None, "qwen_asr": "Polish", "aligner": "Polish", }, "czech": { "paraformer": None, "sensevoice": None, "fun_asr_nano": None, "qwen_asr": "Czech", "aligner": "Czech", }, "filipino": { "paraformer": None, "sensevoice": None, "fun_asr_nano": None, "qwen_asr": "Filipino", "aligner": "Filipino", }, "persian": { "paraformer": None, "sensevoice": None, "fun_asr_nano": None, "qwen_asr": "Persian", "aligner": "Persian", }, "greek": { "paraformer": None, "sensevoice": None, "fun_asr_nano": None, "qwen_asr": "Greek", "aligner": "Greek", }, "romanian": { "paraformer": None, "sensevoice": None, "fun_asr_nano": None, "qwen_asr": "Romanian", "aligner": "Romanian", }, "hungarian": { "paraformer": None, "sensevoice": None, "fun_asr_nano": None, "qwen_asr": "Hungarian", "aligner": "Hungarian", }, "macedonian": { "paraformer": None, "sensevoice": None, "fun_asr_nano": None, "qwen_asr": "Macedonian", "aligner": "Macedonian", }, } # ── 引擎名称常量 ── ENGINE_SENSEVOICE = "sensevoice" ENGINE_QWEN_ASR = "qwen_asr" ENGINE_FUN_ASR_NANO = "fun_asr_nano" ENGINE_PARAFORMER = "paraformer" ENGINE_ALIGNER = "aligner" ALL_ENGINES = [ENGINE_PARAFORMER, ENGINE_SENSEVOICE, ENGINE_FUN_ASR_NANO, ENGINE_QWEN_ASR, ENGINE_ALIGNER] # ── 工具函数 ── def get_language(engine: str, unified_code: str) -> Optional[str]: """ 将统一语言代码转换为指定引擎的标识符。 Args: engine: 引擎名称常量 (ENGINE_*) unified_code: 统一语言代码 (如 "chinese", "english")，不区分大小写 Returns: 引擎特定的语言标识符，若不支持则返回 None """ entry = LANGUAGE_MAP.get(unified_code.lower()) if entry is None: return None return entry.get(engine) def supported_codes(engine: str) -> List[str]: """ 获取指定引擎支持的所有统一语言代码。 Args: engine: 引擎名称常量 Returns: 支持的语言代码列表（按 LANGUAGE_MAP 定义顺序） """ return [code for code, entry in LANGUAGE_MAP.items() if entry.get(engine) is not None] def validate(engine: str, unified_code: str) -> bool: """ 验证指定引擎是否支持该语言代码。 Returns: True 如果支持，False 否则 """ return get_language(engine, unified_code) is not None def list_available() -> Dict[str, List[str]]: """ 列出所有引擎支持的语言代码概览。 Returns: { engine_name: [supported_codes] } """ return {engine: supported_codes(engine) for engine in ALL_ENGINES}