Fix: Replace deprecated cached_path usage with hf_hub_download

README.md

@@ -64,21 +64,28 @@ Usage
 #!pip install transformers==4.20.0
 #!pip install https://github.com/kpu/kenlm/archive/master.zip
 #!pip install pyctcdecode==0.4.0
-from transformers.file_utils import cached_path, hf_bucket_url
+#from transformers.file_utils import cached_path, hf_bucket_url
+from huggingface_hub import hf_hub_download
 from importlib.machinery import SourceFileLoader
 from transformers import Wav2Vec2ProcessorWithLM
-from IPython.lib.display import Audio
+from IPython.display import Audio
 import torchaudio
 import torch
+import kenlm
+
 
 # Load model & processor
 model_name = "nguyenvulebinh/wav2vec2-base-vi-vlsp2020"
 # model_name = "nguyenvulebinh/wav2vec2-large-vi-vlsp2020"
-model = SourceFileLoader("model", cached_path(hf_bucket_url(model_name,filename="model_handling.py"))).load_module().Wav2Vec2ForCTC.from_pretrained(model_name)
-processor = Wav2Vec2ProcessorWithLM.from_pretrained(model_name)
+#model = SourceFileLoader("model", cached_path(hf_bucket_url(model_name,filename="model_handling.py"))).load_module().Wav2Vec2ForCTC.from_pretrained(model_name)
+#processor = Wav2Vec2ProcessorWithLM.from_pretrained(model_name)
+
+model_file = hf_hub_download(model_name, filename="model_handling.py")
+audio_file = hf_hub_download(model_name, filename="t2_0000006682.wav")
 
 # Load an example audio (16k)
-audio, sample_rate = torchaudio.load(cached_path(hf_bucket_url(model_name, filename="t2_0000006682.wav")))
+#audio, sample_rate = torchaudio.load(cached_path(hf_bucket_url(model_name, filename="t2_0000006682.wav")))
+audio, sample_rate = torchaudio.load(audio_file)
 input_data = processor.feature_extractor(audio[0], sampling_rate=16000, return_tensors='pt')
 
 # Infer