Update app.py

app.py

@@ -43,8 +43,10 @@ def infer_tts(ref_audio_orig: str, gen_text: str, speed: float = 1.0, request: g
         gen_texts = chunk_text(gen_text)
         final_wave_total = None
         final_sample_rate = 24000
+        ref_audio, ref_text = "", ""
         for i, gen_text in enumerate(gen_texts):
-            ref_audio, ref_text = preprocess_ref_audio_text(ref_audio_orig, "")
+            if i == 0:
+               ref_audio, ref_text = preprocess_ref_audio_text(ref_audio_orig, "")
             final_wave = generate_sentence(
                ref_text.lower(), 
                ref_audio,