Update inference.py

inference.py

@@ -22,7 +22,10 @@ class InferencePipeline():
         self.trg_spk_emb = torch.from_numpy(self.trg_spk_emb)
         self.trg_spk_emb = self.trg_spk_emb.unsqueeze(0)#.cuda()
 
-    def voice_conversion(self, audio_path):
+    def voice_conversion(self, audio_data):
+        # Extract the file path from the tuple
+        audio_path = audio_data[0] if isinstance(audio_data, tuple) else audio_data
+        
         # load source audio
         source, sr = torchaudio.load(audio_path) #"test.wav")
         source = torchaudio.functional.resample(source, sr, 16000)