bugfix

游雁

2024-07-22 e42f539eb8acd106463a5bc466a39497691fb4ca

 examples/industrial_data_pretraining/sense_voice/demo.py

@@ -7,7 +7,7 @@
from funasr import AutoModel
from funasr.utils.postprocess_utils import rich_transcription_postprocess

model_dir = "iic/SenseVoiceSmall"
model_dir = "/Users/zhifu/Downloads/modelscope_models/SenseVoiceSmall"  # "iic/SenseVoiceSmall"


model = AutoModel(
@@ -19,17 +19,30 @@

# en
res = model.generate(
    input=f"{model.model_path}/example/en.mp3",
    input="/Users/zhifu/Downloads/8_output.wav",
    cache={},
    language="auto",  # "zn", "en", "yue", "ja", "ko", "nospeech"
    use_itn=True,
    batch_size_s=60,
    merge_vad=True,  #
    merge_length_s=15,
    merge_length_s=0.1,
)
text = rich_transcription_postprocess(res[0]["text"])
print(text)

# en
res = model.generate(
    input="/Users/zhifu/Downloads/8_output.wav",
    cache={},
    language="auto",  # "zn", "en", "yue", "ja", "ko", "nospeech"
    use_itn=True,
    batch_size_s=60,
    merge_vad=False,  #
    merge_length_s=15,
)
text = rich_transcription_postprocess(res[0]["text"])
print(text)
raise "exit"
# zh
res = model.generate(
    input=f"{model.model_path}/example/zh.mp3",