Update app.py

app.py

@@ -48,10 +48,18 @@ def transcribe(audio_path):
 
 
     #------
-    audio_input = processor(audio_path, return_tensors="pt", sampling_rate=16000).input_values
+    audio_cnt, sr = librosa.load(audio_path, sr=16000)
+
+    # 将音频数据传递给 processor
+    input_values = processor(audio, sampling_rate=16000, return_tensors="pt").input_values
+    
+    # 模型推理
     with torch.no_grad():
-        predicted_ids = model.generate(audio_input)
+        predicted_ids = model.generate(input_values)
+    
+    # 解码得到转录结果
     transcription = processor.batch_decode(predicted_ids, skip_special_tokens=True)[0]
+    
     return transcription