Update app.py

app.py

@@ -171,21 +171,23 @@ def predict(prompt, language, audio_file_pth, mic_file_path, use_mic, voice_clea
 
         metrics_text= ""
         
-        t_latent=time.time()
-        gpt_cond_latent, _, speaker_embedding = model.get_conditioning_latents(audio_path=speaker_wav)
-        latent_calculation_time = time.time() - t_latent
-        ##metrics_text=f"Embedding calculation time: {latent_calculation_time:.2f} seconds\n"
-        
-        wav_chunks = []
-
-        t_inference=time.time()
-        
-        chunks = model.inference_stream(
-            prompt,
-            language,
-            gpt_cond_latent,
-            speaker_embedding,)            
+                  
         try:
+                t_latent=time.time()
+            gpt_cond_latent, _, speaker_embedding = model.get_conditioning_latents(audio_path=speaker_wav)
+            latent_calculation_time = time.time() - t_latent
+            ##metrics_text=f"Embedding calculation time: {latent_calculation_time:.2f} seconds\n"
+            
+            wav_chunks = []
+    
+            t_inference=time.time()
+            
+            chunks = model.inference_stream(
+                prompt,
+                language,
+                gpt_cond_latent,
+                speaker_embedding,)  
+            
             first_chunk=True
             for i, chunk in enumerate(chunks):
                 if first_chunk: