Dev gzf exp (#1647)

zhifu gao

2024-04-23 2ac38adbe5f4e1374a079e032ed4b504351a207c

 funasr/tokenizer/whisper_tokenizer.py

@@ -1,14 +1,14 @@

try:
   from whisper.tokenizer import get_tokenizer
except:
   print("Notice: If you want to use whisper, please `pip install -U openai-whisper`")

from funasr.register import tables

@tables.register("tokenizer_classes", "WhisperTokenizer")
def WhisperTokenizer(**kwargs):

   try:
      from whisper.tokenizer import get_tokenizer
   except:
      print("Notice: If you want to use whisper, please `pip install -U openai-whisper`")
	
   language = kwargs.get("language", None)
   task = kwargs.get("task", "transcribe")
   is_multilingual = kwargs.get("is_multilingual", True)
@@ -22,3 +22,25 @@
   
   return tokenizer


@tables.register("tokenizer_classes", "SenseVoiceTokenizer")
def SenseVoiceTokenizer(**kwargs):
   try:
      from funasr.models.sense_voice.whisper_lib.tokenizer import get_tokenizer
   except:
      print("Notice: If you want to use whisper, please `pip install -U openai-whisper`")
	
   language = kwargs.get("language", None)
   task = kwargs.get("task", None)
   is_multilingual = kwargs.get("is_multilingual", True)
   num_languages = kwargs.get("num_languages", 8749)
   vocab_path = kwargs.get("vocab_path", None)
   tokenizer = get_tokenizer(
      multilingual=is_multilingual,
      num_languages=num_languages,
      language=language,
      task=task,
      vocab_path=vocab_path,
   )
	
   return tokenizer