from funasr import AutoModel
from funasr.utils.postprocess_utils import rich_transcription_postprocess

model_dir = "./"

# ASR部分用的阿里的开源funasr, 即fun-asr, github地址:
# https://github.com/modelscope/FunASR
model = AutoModel(
    model=model_dir,
    vad_model="fsmn-vad",
    vad_kwargs={"max_single_segment_time": 30000},
    # 注释掉此句的话，可以使用CPU进行识别，不再使用GPU
    # device="cuda:0",
    hub="hf",
)

# en
res = model.generate(
    input=f"{model.model_path}/example/en.mp3",
    cache={},
    language="auto",  # "zn", "en", "yue", "ja", "ko", "nospeech"
    use_itn=True,
    batch_size_s=60,
    merge_vad=True,  #
    merge_length_s=15,
)
# 把转出来的文字做了一个合理化的后处理, 比如去掉一些标点符号, 明显不合理的句子等等.
text = rich_transcription_postprocess(res[0]["text"])
print(text)