update_model_loading

preprocess.py

@@ -125,6 +125,8 @@ def process_audio_file(input_audio_file, num_speakers, device="cuda"):
     print("Loading CrisperWhisper model...")
     device_str = "cuda:0" if torch.cuda.is_available() else "cpu"
     torch_dtype = torch.float16 if torch.cuda.is_available() else torch.float32
+
+    """ Use local Crisper Whisper Model
     local_model_dir = "./CrisperWhisper_local"
 
     cw_model = AutoModelForSpeechSeq2Seq.from_pretrained(
@@ -137,6 +139,22 @@ def process_audio_file(input_audio_file, num_speakers, device="cuda"):
     
     processor = AutoProcessor.from_pretrained(local_model_dir)
     
+    """
+    hf_model_id = "nyrahealth/CrisperWhisper"
+
+    cw_model = AutoModelForSpeechSeq2Seq.from_pretrained(
+        hf_model_id,
+        torch_dtype=torch_dtype,
+        low_cpu_mem_usage=True,
+        use_safetensors=True,
+        token=token
+    )
+    cw_model.to(device_str)
+
+    processor = AutoProcessor.from_pretrained(hf_model_id, token=token)
+
+    
+    
     asr_pipeline = pipeline(
         "automatic-speech-recognition",
         model=cw_model,